# EOMA Documentation — full text > Every page of EOMA's customer documentation, concatenated in reading order. > EOMA (https://eoma.ai) is a marketing intelligence platform that measures and > improves whether a brand is cited in AI search answers. Pages: 40 Individual pages: append .md to any https://eoma.ai/docs/... URL --- # What EOMA is > EOMA measures whether AI assistants mention your brand when buyers ask about your category, and then does the work to change the answer. Source: https://eoma.ai/docs/start-here/what-is-eoma Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - EOMA watches what AI assistants say when people ask about your category, and records whether you were mentioned. - When you are missing, it tells you which specific question you lost and why. - It then produces the work that fixes it — and for many tasks, does that work for you. ## The short version For twenty years, being findable meant ranking on Google. A buyer typed a question, got ten blue links, and clicked one. Every marketing tool ever built was pointed at that moment. That moment is being replaced. A buyer now asks ChatGPT, Claude, Perplexity, Gemini, or Google's own AI summary — and gets *one answer* instead of ten links. That answer names two or three brands. If you are not among them, you were not beaten to position four. You were not in the conversation at all, and the buyer never learned you exist. EOMA is built for that moment. It asks the AI engines the questions your buyers actually ask, reads the answers, and records what happened: were you mentioned, which competitors were, which websites the engine trusted enough to quote. Then it turns that into work — the content, the pages, the corrections that make the next answer include you. The distinction worth holding on to: traditional SEO tools tell you where you *rank*. EOMA tells you whether you are *in the answer*. Those are no longer the same question, and a brand can do well at one while being invisible at the other. ## How it works 1. **We build your question set.** From your website, your category, and your market, EOMA assembles the questions a real buyer would ask an assistant about what you sell. Not keywords — whole questions, in the language your buyers use. 2. **We ask the engines, on a schedule.** Every question goes to each AI engine repeatedly, week after week. One answer is an anecdote; the same question asked over months is a measurement, because these engines do not answer identically twice. 3. **We read every answer.** For each one: were you named, which competitors were named, which sources did the engine cite, and did it cite *you* or merely repeat something you published elsewhere. That last distinction matters more than it sounds — see [How your visibility is scored](/docs/concepts/how-visibility-is-scored). 4. **We turn the gaps into work.** A question you keep losing becomes a specific, scoped task: this page needs writing, this claim needs a source, this profile needs correcting. Most weeks EOMA can carry out the task itself and hand you the result to approve. 5. **We measure again next week.** Because the answer changes, and the only proof a fix worked is the engine's behaviour afterwards. ## What this means for you **Your numbers start small and get more meaningful every week.** EOMA accumulates. In week one you have a snapshot; by week eight you have a trend, and the trend is what actually tells you whether you are winning. A low score in your first fortnight is a starting line, not a diagnosis. **You will see competitors you did not consider competitors.** The engines cite whoever answered the question well — a directory, a forum thread, a newspaper, a rival three times your size. For the purpose of one question, that is who you are competing with, and pretending otherwise costs you the answer. **Being invisible is usually fixable, and usually not with more advertising.** The engines quote sources they can read, verify, and trust. That is a content and evidence problem far more often than a budget problem, which is why EOMA's output is work rather than a bid strategy. **You do not need a technical team.** Connecting your accounts is a series of "sign in with Google" style approvals, and the work EOMA produces arrives as drafts you approve, not as code you deploy. ## Common misunderstandings | People assume | Actually | |---|---| | "This is SEO with a new name." | SEO optimises your position in a list of links. EOMA optimises whether you appear in a generated answer that may show no links at all. Both matter; they are measured differently and fixed differently. | | "If I rank first on Google, AI will mention me." | Frequently untrue. The engines draw on sources they were trained on and sources they can retrieve now. Brands that rank well and are still absent from AI answers are the ordinary case, not the exception. | | "One bad score means the tool is wrong." | Ask the same question twice and these engines answer differently. That is why EOMA asks repeatedly over time. Judge the trend, not a single run. | | "The score should improve as soon as I publish something." | The engines re-read the web on their own schedule. Weeks, not hours, is the normal delay between publishing a fix and seeing the answer change. | | "EOMA writes the content and that is the whole product." | The writing is the visible part. The measurement is what tells you which writing was worth doing, and whether it worked. | ## Related - [How your visibility is scored](/docs/concepts/how-visibility-is-scored) — what the number on your dashboard is made of - [Your first week](/docs/start-here/your-first-week) — what to expect, day by day - [AI provider visibility](/docs/pages/ai-provider-visibility) — the screen where the measurement lives --- # Your first week > What happens day by day after you sign up, what will look empty or wrong early on, and why most of that is normal. Source: https://eoma.ai/docs/start-here/your-first-week Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-09-21 --- ## In 10 seconds - Day one is setup and your first report. Most other screens will be thin, and that is expected. - Your second measurement lands around day four, which is the first point any number means anything. - The one thing worth doing in week one that pays off forever: check that your question set actually asks about your business. ## The short version EOMA accumulates. Nearly every number it shows is a pool that grows every week, so the day you sign up is the day your data is at its thinnest it will ever be. A first-week score is a starting line, not a diagnosis. That has a practical consequence: half the "something is broken" moments in week one are screens correctly reporting that they have nothing yet. This page tells you which ones those are, so you spend your first week on the things that compound instead of on the things that will fill themselves in. The work that compounds is all in the same place: **the facts you give EOMA about your business.** Your website, your real competitors, your markets, your industry. Those are what your questions are written from, and the questions are what everything else is downstream of. Twenty minutes there is worth more than anything else you can do in week one. ## How it works **Day 1 — signup and setup.** After the card, you go through seven short screens: your brand basics, your website, your competitors, your markets, your industry, your team size, and how you found us. These are pre-filled from a first read of your site, and they are a **review of what was found**, not a form. If a pre-filled answer is wrong, correct it now — an industry left on a placeholder value ends up shaping your entire question set. **Day 1, minutes later — the build.** A progress card takes over your dashboard while the pipeline runs. It fills in five sections as they complete: your brand snapshot, competitors found, social signals, AI-search visibility, and share of voice. This keeps running whether or not you watch it, and you can leave and come back. **Day 1 — your first visibility report.** Your first real numbers. Treat this as the "before" photograph. If your visibility percentage is low, that is the ordinary case and it is the reason you bought the product. **Day 1–2 — the Start here checklist.** Four deep links on your dashboard: review your first visibility report, see your share of voice against competitors, open your action board, and connect your site so fixes publish automatically. It disappears when you finish or dismiss it. **Day 2–3 — read your question set.** This is the highest-value hour of your first week. Open your prompts and read what EOMA is actually asking on your behalf. Are the questions ones your buyers would type? Is your second product line represented? Is the market right? Fix them now, before you have weeks of history measuring the wrong thing. **Day 4–5 — connect a publishing surface.** WordPress or GitHub. This is what turns a recommendation into a change on your site instead of a draft you have to copy out by hand. It is optional — you can always export a draft and apply it yourself — but everything downstream is faster with it. **Day 5–7 — start one action.** Not five. Start the single most important row and watch it go all the way through to something published. You will learn more about how EOMA works from one completed action than from any amount of reading. **Day 7 — your second run.** Paid plans measure once a week. This is the first moment you have two points instead of one. They will disagree, and that disagreement is information about how unstable the engines are in your category, not a fault. ## What this means for you **Screens that will be empty or thin in week one, and why.** None of these are faults: | What you see | Why | |---|---| | Sentiment reads "not measured" | Tone analysis is a separate paid step and has not been run for your brand. It is deliberately blank rather than showing a neutral score nobody measured. | | Your trend chart has one point, or none | A trend needs several windows. Expect a usable line from around week three. | | Rank tracking is empty | Your keyword set has to be seeded before weekly search-results checks begin. If it is still empty after a few days, tell us — it is not always automatic. | | Social listening shows nothing | It is Growth and above. On Growth, zero X mentions after a week is worth reporting; on Starter, the section is locked by plan. | | The sources board shows very few citations of your own site | Common and correct. Most brands have very few citations they earned from questions that did not name them. | | Some competitors you have never heard of | The engines cite whoever answered the question well, including directories and marketplaces. | **Your numbers will move in the first fortnight for reasons that are not about you.** The 28-day scoring window is still filling. Judge nothing before week three, and judge the direction rather than the level. **A low first score is the normal starting point.** Most brands are absent from most answers when they arrive. That is the gap the product exists to close. **What is genuinely worth reporting in week one:** a screen that spins forever, a brand analysis that failed and will not retry, an AI Visibility page that is still completely empty after your second run, or data that has not moved in over a week. Those are real. ## Common misunderstandings | People assume | Actually | |---|---| | "Setup is a form, so I can click through it." | It is a review of what was read off your site, and the answers become your question set. A pre-filled industry left uncorrected shapes months of measurement. | | "The build failed because it is taking a while." | The pipeline runs several collections and keeps going after you navigate away. A genuine failure shows an explicit failure card with a retry. | | "My score is low, so the tool is measuring the wrong thing." | Possibly — check your question set. But low is also the ordinary honest answer for a brand the engines have not learned yet. | | "I should start every recommended action in week one." | On Growth, EOMA runs three jobs at a time and a seat frees as each finishes; on Starter you have three executions in total. Start the one that matters most and learn the loop. On Growth the list tops itself up every week; on Starter each fixed task comes back on its own schedule. | | "Week one numbers are my baseline." | Week three is your baseline. Week one is a snapshot taken before the window filled. | ## Related - [What EOMA is](/docs/start-here/what-is-eoma) — the product in one page - [What EOMA does, and what you do](/docs/start-here/what-eoma-does-and-what-you-do) — the honest division of labour - [Why your prompts matter](/docs/concepts/why-prompts-matter) — the hour that pays off most --- # The vocabulary > Eleven words EOMA uses constantly — AEO, engine, prompt, run, citation, source, share of voice, gap, action, brand index — each defined in one line. Source: https://eoma.ai/docs/start-here/the-vocabulary Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Eleven words carry most of EOMA. Learn these and every screen becomes readable. - The two worth learning first are **prompt** (a question EOMA asks on your behalf) and **run** (one round of those questions going out). - Each definition below links to the page that goes deeper. ## The short version EOMA measures something the industry has not agreed on a name for yet, so its vocabulary is a mixture of borrowed search terms and new ones. Most of the confusion is in three places: *prompt* means a question and not an instruction to a chatbot, *source* and *citation* are related but not the same, and *engine* means an AI assistant rather than a search engine. The rest is ordinary. Read this once and keep it as a reference. ## How it works **AEO — Answer Engine Optimisation.** Making your brand more likely to appear in the answer an AI assistant generates. The equivalent of SEO for a world where the buyer gets one answer instead of ten links. → [What EOMA is](/docs/start-here/what-is-eoma) **GEO — Generative Engine Optimisation.** The same idea under a different name. Different vendors picked different letters; EOMA treats them as synonyms and uses AEO. **Engine.** One AI assistant. EOMA asks four: ChatGPT, Claude, Gemini and Perplexity. They are asked the same questions and they frequently disagree, which is itself a finding. → [AI provider visibility](/docs/pages/ai-provider-visibility) **Prompt.** One whole question EOMA puts to the engines on your behalf — "which sites are best for renting a villa in Antalya?" — not a keyword and not an instruction. Your set is written from your brand profile, and you can edit it, switch questions off, and add your own in any language. → [Why your prompts matter](/docs/concepts/why-prompts-matter) **Run.** One round of your questions going out to the engines and the answers being recorded. Paid plans run once a week. Each run takes a quarter of your question set, so four runs cover all of it. → [What a run is](/docs/concepts/what-a-run-is) **Citation.** A source the engine quoted or linked in an answer. The distinction that matters: a citation of your own site inside an answer to a question that *named* your brand is your own echo, not evidence anyone found you. EOMA ranks on citations you **earned** — from questions that never mentioned you. → [How your visibility is scored](/docs/concepts/how-visibility-is-scored) **Source.** The website a citation points at — your domain, a competitor's, a directory, a newspaper. The sources board ranks the domains the engines trust enough to quote in your category, with your own row marked in place. → [Your brand on Google](/docs/pages/your-brand-on-google) **Share of voice.** Your slice of every brand mention across your answers. Two views: the brands the engines named on their own, and the competitor list you keep. They will disagree, and both are right. → [Competitors and share of voice](/docs/concepts/competitors-and-share-of-voice) **Gap.** A question you keep losing — the engines answer it, and you are not in the answer. Gaps are where the recommended work comes from, because a gap is a specific, addressable thing rather than a general low score. You meet them as the to-dos on your Actions board rather than as a list of their own. → [Actions](/docs/pages/actions) **Action.** One specific job on your board: write this page, add this source, correct this profile. Each one is either something EOMA can do for you or something only you can do, and the board says which. → [The life of an action](/docs/concepts/life-of-an-action) **Brand index.** The canonical record of what is true about your business — name, contact details, languages, reach — as the engines should have it. You confirm and correct it, and it is what EOMA uses when a correction needs to be published somewhere. → [Brand index](/docs/pages/brand-index) ## What this means for you **When a screen says "visibility", ask which one.** EOMA has two related figures that are easy to conflate: the rate at which you appear in answers, and the eight-axis score that weights being *suggested* far above being *known*. They describe the same reality with different denominators and they will not match. **"Mentions" and "answers" are different denominators.** A single answer that names a competitor twice contributes two mentions. Percentages computed on mentions and percentages computed on answers can differ several-fold while both being correct. **"Tracked" and "held" are different for keywords too.** Your pool is what you hold; your tracked set is the much smaller group actually checked each week. → [Plans and limits](/docs/start-here/plans-and-limits) **"Active" means eligible, not scheduled.** An active prompt is a candidate for the next run, not a promise it will be asked this month. ## Common misunderstandings | People assume | Actually | |---|---| | "A prompt is what I type into ChatGPT." | Here it is a question EOMA asks *on your behalf*, as a buyer would phrase it, and records the answer to. You do not chat with it. | | "Engine means Google." | It means an AI assistant. Google appears in EOMA under its own heading, because Google's AI answer is a separate thing from ChatGPT's. | | "Citations and sources are the same word." | A citation is one act of quoting; a source is the domain quoted. Ten citations of one newspaper is one source. | | "AEO and GEO are two products." | Two names for the same discipline. EOMA uses AEO throughout. | | "A gap is a bug report." | A gap is a market opportunity: a question being answered without you in the answer. | ## Related - [What EOMA is](/docs/start-here/what-is-eoma) — the product in one page - [How your visibility is scored](/docs/concepts/how-visibility-is-scored) — the number these words add up to - [Dashboard](/docs/pages/dashboard) — where most of these appear together --- # Plans and limits > What the free plan, Starter, Growth and Scale include, and which numbers are the ones that actually bind. Source: https://eoma.ai/docs/start-here/plans-and-limits Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-07 --- ## In 10 seconds - Three paid plans: Starter $49/month, Growth $99/month, Scale $249/month. Billing is monthly or yearly; a year costs ten months (two months free). - The free plan asks for no card; a paid plan is charged the day you choose it, and you get the plan you chose. - The limits that bind in practice are tracked questions, tracked keywords, and how many actions EOMA does for you. ## The short version EOMA prices on how much measurement you want and how much work you want done for you. Everything else (the reports, the integrations, the onboarding, the action board) is on every plan, though what lands on the board grows with the plan. One integration is the exception: Google Search Console starts at Starter and is not on the Free plan. Google Analytics and the five publishing connections (WordPress, HubSpot, Wix, GitHub, Vercel) are on every plan, Free included. | | **Starter** $49/mo | **Growth** $99/mo | **Scale** $249/mo | |---|---|---|---| | Brands | 1 | 1 | 1 | | Markets tracked | 1 | 1 | 1 | | AI questions tracked per month | 30 | 100 | 200 | | Keywords tracked | 200 | 2,000 | 4,000 | | Actions on your board | 6 fixed tasks, each on its own schedule | Starter's 6 + 3 more fixed checks + new actions every week from your measurements | Starter's 6 + 3 more fixed checks + new actions every week from your measurements | | Done-for-you executions | 3 in total | 3 at a time | 6 at a time | | Support | — | email | priority | Larger scopes are agreed with the team — contact `support@eoma.ai`. Three surfaces are Growth and above: **social listening**, the deep AI-citation analysis (**source map, freshness and opportunity gaps**), and **Your Brand on Google**. Starter is a paying plan that sees a lock on those, deliberately. ## How it works 1. **The free plan needs no card.** You sign up, verify your email, and the first measurement runs on the free plan: 16 questions a month on ChatGPT, refreshed every two weeks, 100 keywords (10 of them checked in Google every month), and two jobs EOMA does for you in total. Moving to a paid plan charges the card that day; the plan you pick is the plan you get. 2. **Question limits cap how many are *active*, not how many you own.** Your plan's number is the ceiling on active questions: 16 on the free plan, 30 on Starter, 100 on Growth, 200 on Scale. Switching one off frees a slot, and a switched-off question keeps every answer it ever earned. Your own hand-written questions have a separate pool of 100 per brand, the same on every plan, because that is a size limit rather than something you are being sold. 3. **Keywords have two different caps and both are visible.** The **pool** cap is how many keywords your brand may hold: 100 on the free plan, 200 on Starter, 2,000 on Growth, 4,000 on Scale. The **tracked** cap is how many get a full search-results check: 10 on the free plan and 25 on Starter, checked every month; 50 on Growth and 100 on Scale, checked every week. That second number is also the cap on how many keywords you may pin, because a pin spends from the same budget rather than adding to it. The portfolio screen quotes it back to you as "N of 50 pinned". 4. **Done-for-you executions are seats on Growth and Scale, and a fixed total on Starter.** Growth's limit is a concurrency cap and nothing else: three done-for-you jobs may run at the same time, and a seat frees the moment one of them reaches done or you remove it. Scale holds six. There is no weekly quota to run out of and nothing refills on Monday — over a quiet month you can start as many as you like, three at a time. Starter gets three in total: they do not renew, and once they are spent EOMA does no further done-for-you work on that account. When the meter reads 3 / 3, every seat is busy. One exception, for a short while: work that was already in flight before 2026-09-08 does not take a seat, so you may see more rows in progress than you have seats. That is expected, and it ends as those jobs finish. 5. **The board is not metered.** On Starter it holds six fixed tasks, each coming back on its own schedule; Growth and Scale add three more checks and new actions every week from what EOMA measured. What is metered is EOMA doing the work for you. You can work any row yourself and mark it done at no cost. 6. **Your AI measurement refreshes every week** on every paid plan — four runs a month, each asking a quarter of your questions, so your monthly question budget is spread rather than spent at once and every question you track is asked once a month. On the free plan it refreshes every two weeks. Search-results tracking refreshes every week on Growth and Scale and every month on Starter and the free plan; social listening, on Growth and Scale, refreshes every week. ## What this means for you **Pick your plan on questions and listening, not on price.** 100 questions is a real set for a single-market business with one product line. If you sell in three countries or have several distinct product lines, 100 questions will each be covering too much ground and your numbers will be thin. That is the reason to be on Growth, more than the social listening. **If you are on Starter or the free plan and a screen shows a lock, that is intentional.** Social listening, the AI source map, freshness, opportunity gaps and Your Brand on Google are Growth features. **Starter's three executions do not renew — spend them on what matters most.** On Starter, EOMA carries out three actions for you in total, not three at a time. Plan to spend them on the highest-value items and do the rest yourself: every action's detail copies its instructions as a prompt for your own AI, and the board itself is not metered. Growth's three seats are what turn done-for-you work into a running habit: as one job finishes, the next one starts. **Watch the tracked-keyword number, not the pool number.** The pool is how many you may hold; the tracked set is how many are actually being checked, and it is much smaller. A keyword in your pool but not in your tracked set is not being measured. **A year costs ten months.** If you plan to stay for a year, paying yearly is the cheapest form of the same plan. ## Common misunderstandings | People assume | Actually | |---|---| | "200 questions means 200 AI calls a month." | One question produces seven answers — ChatGPT, Claude and Gemini are each asked twice (once allowed to search the web, once not), and Perplexity, which is search-native, once. The engines and the modes are delivery, not units of value — the thing counted against your plan is the distinct question. | | "I have 2,000 keywords on Growth, so 2,000 are being checked." | 50 are checked weekly. The other 1,950 are in your pool, available to promote or pin, and are not consuming a weekly search-results check. | | "I did not use my executions, so they stack up." | There is nothing to stack. On Growth and Scale, executions are seats, not a weekly quota — three (or six) jobs can be in flight at once, and an idle seat is an open seat, waiting for the next Start. On Starter the three executions neither renew nor accumulate. | | "I can add every country I sell into." | Your plan measures one market, and that market decides two things: where your visibility is measured, and what language your tracked questions are written in. A market is the country whose **buyers** you serve — not where you are, and not what you talk about. Several markets is an Enterprise conversation. | | "Upgrading gives me back the data I could not see." | Upgrading unlocks the surface and starts the collection going forward. Data that was never collected on the lower plan does not appear retroactively. | ## Related - [Plan and usage](/docs/settings/plan-and-usage) — your current plan, your usage, and where to change it - [The life of an action](/docs/concepts/life-of-an-action) — what a done-for-you execution actually buys - [Why your prompts matter](/docs/concepts/why-prompts-matter) — how to spend your question allowance well --- # What EOMA does, and what you do > An honest division of labour — what runs without you, what needs your approval, and what EOMA cannot do at all. Source: https://eoma.ai/docs/start-here/what-eoma-does-and-what-you-do Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-09-21 --- ## In 10 seconds - Measurement is fully automatic. It runs on a schedule and needs nothing from you after setup. - The work is produced for you, but publishing it is a decision you make, every time. - Some things EOMA cannot do at all — and the board says so rather than pretending. ## The short version EOMA is closer to "does the work" than "gives you a dashboard", but the honest answer has a boundary in it and the boundary is worth understanding before you buy. **Automatic, forever, without you:** asking the engines, reading the answers, scoring them, tracking competitors, finding the gaps, and keeping your board of what to do current (on Growth and above, with new actions written every week from those gaps). You could ignore EOMA for a month and come back to a full history and a current board. **Automatic once you say go:** producing the actual work. Press Start on an action and an agent researches it, drafts it, and brings it back finished. Most weeks that is a real page, a schema block, a correction — not an outline. **Always yours:** deciding whether it goes live. EOMA will not publish to your site without you approving that specific piece and choosing the destination. On a repository it opens a pull request and stops; merging is yours. **Never EOMA's:** anything that needs a person in the world. Calling a supplier, getting listed in a trade directory that requires a phone verification, changing your prices, getting a journalist to write about you. Those appear on your board under a header that says EOMA cannot do them, with the steps written out. The rough shape: EOMA removes the *finding out* and most of the *making*, and leaves you the *deciding* and the *doing things in the world*. ## How it works **What runs with no input from you at all** | | How often | |---|---| | Asking your questions to four AI engines and recording the answers | Every 3 days on paid plans | | Scoring those answers across eight axes | After every collection | | Folding competitor names, tracking who the engines cite | Continuous | | Weekly search-results checks on your tracked keywords | Weekly | | Social listening on X and LinkedIn (Growth and above) | Weekly | | Writing new actions from your measurements and topping the list up (Growth and above) | Weekly | | Emailing you when something needs your attention | As it happens | **What needs one click from you** - **Start** on an action — spends one done-for-you execution, irreversible. - **Answering a question** the agent asks when it hits a fact only you have. - **Approving a draft** and choosing where it goes: live on your site, or as a pull request you review line by line. - **Merging** a pull request, if that is the destination you chose. **What is entirely yours** - **Setup accuracy.** Your website, competitors, markets and industry. These are what your question set is written from, and nothing downstream can correct a wrong answer here. - **Reading your question set once and fixing it.** EOMA writes it; only you know whether it covers your second product line or your second market. - **Connecting your accounts.** Google Analytics, Search Console, your CMS or repository. Each one is a sign-in style approval, not a configuration project. - **Actions EOMA cannot ship.** Anything requiring a phone call, a signature, a relationship, or a change to something EOMA has no access to. - **Judgement about your own business.** EOMA will tell you a directory out-ranks you on a buying question. Whether that is worth your week is your call. ## What this means for you **You do not need an engineer, and you do not need a marketing team.** The connections are sign-in approvals; the output arrives as drafts to approve rather than code to deploy. Almost every EOMA customer is a founder without an in-house marketing function — that is who the product is shaped for. **But you do need about an hour in week one.** Setup, reading your question set, and connecting one publishing surface. Skipping the second of those is the single most common way people end up with months of measurement pointed at a market adjacent to their own. **And you need roughly one decision a week after that.** Your plan lets EOMA run three done-for-you jobs at a time on Growth, and a seat frees as each one finishes; on Starter it gives you three in total, which do not renew. Realistically that is a weekly appointment of fifteen minutes: read the board, start the one that matters, approve last week's draft — on Starter, the starting part happens three times in all. **Do not expect EOMA to close the loop on your repository yet.** When you merge a pull request, nothing in EOMA notices automatically — the row stays In progress until someone closes it. This is a known gap being worked on, and it is the one place where the hand-off is genuinely rougher than it should be. **The measurement is worth something even if you never start an action.** Knowing which questions you lose, and to whom, is a real input into decisions EOMA has nothing to do with — pricing, positioning, which market to enter next. ## Common misunderstandings | People assume | Actually | |---|---| | "This is an analytics product, so I have to interpret it myself." | The recommendation list is the interpretation. The screens exist so you can check the reasoning, not so you can do the reasoning. | | "It is an autopilot, so I can leave it running." | Measurement runs unattended; work does not. Nothing is produced until you press Start, and nothing is published until you approve. | | "It will publish things to my site without asking." | It cannot. Approval names the specific piece and the specific destination, and the live-site path asks you to confirm that visitors will see it immediately. | | "If EOMA cannot publish it, the work is wasted." | Every draft exports as Markdown, HTML or PDF, or copies as a prompt. You can apply it by hand and mark the row done. | | "The recommendations are generic content-marketing advice." | On Growth and above, the weekly ones are generated from your own measured gaps: a specific question you lost, to a specific competitor, on a specific engine. Starter's board holds a fixed set of six routines instead. | | "I can hand this to an agency and stop thinking about it." | You can, and several customers do. The setup accuracy and the weekly approval still need somebody who knows the business. | ## Related - [Your first week](/docs/start-here/your-first-week) — what that hour actually looks like - [The life of an action](/docs/concepts/life-of-an-action) — the approval loop in detail - [Plans and limits](/docs/start-here/plans-and-limits) — how much work per week you are buying --- # Measuring AI traffic > What to set up so visits from ChatGPT and other assistants show up in your own analytics — and the two kinds of AI visit nobody can count. Source: https://eoma.ai/docs/start-here/measuring-ai-traffic Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-16 --- ## In 10 seconds - Some AI visits leave a trace in your analytics and some leave none, so the number you can count is a **floor** — the real figure is higher, always. - There are four things to set up. One takes a click, one takes two toggles, one depends on where your website is hosted, and one is a question on your contact form. - Very few AI mentions turn into clicks. A small number is the normal state of this channel, not a sign the work failed. ## The short version An AI assistant can talk about your business in four different ways, and only the last one is countable. It can **describe you with no link at all** — the person reads the answer and never leaves the chat. It can **show a link nobody clicks**. It can **send someone to your site in a way that hides where they came from**. Or it can send someone in a way that says plainly: this visit came from ChatGPT. Your analytics only ever sees that fourth case. Everything before it happened out of sight. What makes the fourth case work is a small note the browser passes along, called the **referrer** — "the person was on chatgpt.com before landing here". ChatGPT also tags its links with a marker of its own (`utm_source=chatgpt.com`), which is useful because that marker survives even when the referrer note is missing. EOMA reads your Google Analytics live and checks each traffic source against its own list of AI assistants. Where one matches, the source gets an **AI** badge on **Where Visitors Came From**, and the panel adds a line above the list: *"41 sessions came from AI assistants"*. Sources with no referrer at all are grouped as **Direct / no referrer** — and some unknown share of that group is AI traffic that arrived without its note. ## How it works 1. **Someone asks a question.** The assistant fetches your page to read it. This visit is a robot, not a person, so it never appears in analytics — it appears only in your website's server records. 2. **The assistant answers,** sometimes naming you, sometimes with a link, sometimes with neither. 3. **If the person clicks,** their browser usually passes the referrer note, and Google Analytics records the source as `chatgpt.com`, `perplexity.ai`, and so on. 4. **If they clicked inside a phone app,** the note is dropped before it reaches you. This is how iPhone and Android open links inside apps; no setting on your side changes it. That visit lands in **Direct / no referrer** instead. 5. **EOMA labels what arrived** each time you open the screen. Nothing is imported overnight — the figures are Google's own, read at that moment. The four things worth setting up, in the order they pay off: | | What | Effort | |---|---|---| | 1 | **Connect Google Analytics** to EOMA — see [Google Analytics 4](/docs/integrations/google-analytics) | one sign-in | | 2 | In Google Analytics, open Admin → Data Settings and turn **Google Signals off**, or set Reporting Identity to **Device-based** | two toggles | | 3 | If your site can sit behind Cloudflare, put it there — the free plan includes a report of which AI robots read which pages | a DNS change | | 4 | Add a **"How did you hear about us?"** box to your contact or quote form — free text, not a dropdown | five minutes | Step 2 is the one people skip and the one that costs most. Google hides any row in a report covering fewer than about fifty people, to keep individuals from being identifiable. If fifteen people arrived from ChatGPT last month, that row is not shown as fifteen — it is not shown at all, and the screen reads zero. Those two toggles are what stop a real number from being reported as nothing. Step 3 is the only way to see the robot half of the story — the assistants reading your pages, which happens far more often than anyone clicks. If your site runs on Shopify, Wix or Squarespace, this step is not available to you; those platforms do not expose their records to anyone, including us. Step 4 catches what no software can. Someone reads about you in the ChatGPT app, types your name into their browser the next morning, and fills in your form. There is no trace to follow. Asking them is the only method that works, and a free-text box beats a dropdown because people write "chatgpt" in words you would never have thought to list. ## What this means for you **Treat the number as a floor and say so out loud.** "At least 40 visits came from AI assistants last month" is a sentence you can defend. "40 visits came from AI" is not, because the app clicks and the copy-paste visits are missing from it and always will be. **Do not read a low click count as failure.** Assistants read pages constantly and cite a fraction of what they read; of the pages they cite, only a fraction get clicked. A business can be recommended regularly and see a handful of visits. That is the shape of this channel today. **Keep some pages untouched.** If every page changes at once, and AI traffic then rises, you cannot tell your work apart from the whole market growing — and it is growing quickly on its own. Leaving a set of pages alone gives you something to compare against, and it has to be decided before the work starts, not after. **Two toggles are worth more than any tool.** Step 2 takes a minute and decides whether the rest of this is visible at all. ## Common misunderstandings | People assume | Actually | |---|---| | "Nothing shows in analytics, so AI is not recommending me." | Most mentions never become clicks, and a share of the clicks arrive with no trace. Absence of traffic is not absence of recommendation. | | "The number in EOMA is the total." | It is the part that can be proven. The untraceable part is real and uncounted. | | "Google's AI answers should show up as AI traffic." | Google files clicks from its AI answers as ordinary search, and gives nobody a way to separate them. Those visits sit in your organic search figures. | | "Adding a tracking script to my site would fix it." | A script reads the same missing note. Where the referrer was dropped before the visit arrived, there is nothing on the page for any script to find. | | "Zero means zero." | Google hides small rows. Until step 2 is done, zero can mean "too few people to display". | ## Related - [Google Analytics 4](/docs/integrations/google-analytics) — the connection steps 1 and 2 depend on - [Web analytics](/docs/pages/web-analytics) — where the AI badge and the sessions line appear - [What EOMA does, and what you do](/docs/start-here/what-eoma-does-and-what-you-do) — the split this page is an instance of --- # Dashboard > The screen you land on. One sentence saying where your brand stands with the AIs, the three numbers behind it, and the actions prepared for you this week. Source: https://eoma.ai/docs/pages/dashboard Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-07 --- ## In 10 seconds - The sentence at the top is the answer. Everything below it is the evidence for that sentence. - The **Actions** card is the only part that asks anything of you. Its sentence says what waits for you, in your plan's words. The rest is reading. - Nothing here needs setting up. It fills in on its own as each weekly scan finishes. ## Before you start Nothing — this page works from your first run. It is the first screen after you sign in, and the EOMA mark in the top-left corner returns you to it. If your brand was created in the last hour you will see a progress card instead of the numbers. That is timing, not a fault, and it is covered in "When it goes wrong". ## Steps ### 1 — Read the sentence at the top One line, in plain words, that says where you stand: *"The AIs are not recommending you yet"*, or *"You are on a good path, the gap to your competitors is closing"*, and so on. It changes as your score does. Read it before you look at any number. ``` ┌─ Panel ───────────────────────────────────┐ │ Your brand's week, and the actions … │ │ │ │ ► The AIs are starting to see you. │ ← read this first │ │ │ ┌────────┐ ┌────────┐ ┌────────┐ │ │ │ 27% │ │ 25.3 │ │ 513 │ │ ← then these three │ └────────┘ └────────┘ └────────┘ │ └───────────────────────────────────────────┘ ``` ### 2 — Read the Actions card The card named **Actions** shows the road every action walks, as three numbers, and one sentence under them that says what to do next. ``` ┌─ Actions ──────────────────────────────────────────┐ │ Waiting for you Waiting for your Completed │ │ to start 2 → review 0 → 0 │ ← where your actions stand │ │ │ EOMA has prepared 2 actions. On your Free plan │ ← what to do, in your │ EOMA does 2 jobs for you; just press Start. … │ plan's words │ [ Start your actions ]│ ← then this └────────────────────────────────────────────────────┘ ``` ### 3 — Press the card's button The button at the foot of the card opens the actions board, and its words say what is waiting there: **Start your actions** when actions are waiting for you to start them, **Review 2 actions** when drafts or questions are waiting for you, and **See all actions** otherwise. Pressing it starts nothing and uses nothing — only **Start** on the board hands an action to EOMA. See [Actions](/docs/pages/actions). ## What you'll see ### The header — "Panel" The tool name, your brand beside it, and one line: *"Your brand's week, and the actions we prepared for you."* If no brand is selected it reads "Your brand". ### The sentence, and its button Five versions exist and which one you get is decided by your AI visibility score, not chosen by hand. Below 10 it says the AIs are not recommending you yet; below 25, that they are starting to see you; below 45, that the gap is closing; at 45 and above, that visibility is strong and the job is growing it. Before your first scan it says your first score lands with that scan. Under it, one line naming how often your brand comes up in the answers we read. ### The three numbers Left to right: **When nobody says your name**, **AI visibility score**, **AI answers analyzed**. Each carries its own window in small type underneath — "all time", "last 8 scans" — so no figure has to be assumed. A number with no data yet reads "—" and says *Fills with the first scan*; it never shows a zero it cannot support. ### "Actions" Three numbers in the order an action walks them — **Waiting for you to start**, **Waiting for your review**, **Completed** — named as the board's own tabs are named, so the number here is the number there. Under them, one sentence in your plan's words. On the Free plan with two fresh actions it reads *"EOMA has prepared 2 actions. On your Free plan EOMA does 2 jobs for you; just press Start. You can also do them yourself."* On Growth it says EOMA runs 3 of your jobs at a time instead. While something waits for you, the number that needs you is lit, a coloured light runs around the card's edge, and the card's button is the page's main button. Review comes first: when a draft waits for your approval, that number is lit even if there are also actions to start. When nothing waits for you, the sentence says what EOMA is doing — *"EOMA is working on 1 of your actions. The draft comes to you for review when it is ready."* — or when the next actions arrive. Before your first scan finishes the card has no numbers yet. It reads *"Your first actions arrive with the first scan"* instead. ### "Where your brand shows up" Four tiles — **AI Visibility**, **SEO**, **Social**, **Analytics** — each a link to that area, each with one line saying what it answers. A fifth tile, **See your AI traffic**, reports how much of your Google Analytics setup is connected: not started, some sources live, or both live. ### "Our weekly rhythm for you" A footer strip with the last scan and the next one. It exists to make one thing plain: *"It runs by itself every week. Your only job is approving the actions."* ## Reading the numbers **When nobody says your name — for example 27%.** This is the one you are buying. Of the AI answers we read where your brand was never mentioned in the question, 27% still named you. Someone who has never heard of you asked about your category, and roughly one answer in four brought you up. The line underneath gives the raw count — "named in 138 of 511 answers · all time" — so you can check the figure rather than trust it. **AI visibility score — for example 25.3.** Your overall standing, on a scale of 0 to 100. It is an index, not a percentage, which is why it carries no `%` sign: half of it is scored against the strongest brand in your own market rather than against a perfect 100. So 25.3 is a position among real rivals, not "a quarter of the way there" — [AI provider visibility](/docs/pages/ai-provider-visibility) gives the range real leaders reach. A small arrow beside it shows the change since the previous scan. The same number, split into its two halves, opens [AI provider visibility](/docs/pages/ai-provider-visibility). **AI answers analyzed — for example 513.** How many AI answers EOMA has read about your category since you started, across every engine, with the number of scans underneath. It is a running total, so it only ever climbs. It matters because it is the denominator: the first two numbers are only as trustworthy as the pile they were measured on, and a brand three weeks in has a smaller pile than one six months in. **Waiting for you to start — for example 2.** Actions EOMA has prepared for your brand that nobody has started. Nothing happens to them until you press **Start** on the board, or do them yourself with the steps inside each one. Free and Starter give EOMA a set number of jobs for the life of the plan, and the sentence says how many are left. On Starter, when they are used, this number can still be above zero: those actions are yours to do, or a reason to upgrade. On Free, once both jobs are used, the other cards wait behind a lock on the board and this number is 0. **Waiting for your review — for example 1.** Drafts waiting for your approval plus actions where EOMA needs information from you. These have stopped and only you can move them; nothing is published until you approve it. **Completed — for example 4.** Every action finished for your brand, the same number the board's **Completed** tab shows. It is a running total, not this week's count. **The number on the AI Visibility tile.** The same visibility score as above, repeated on the tile so you can see it while deciding where to click. **The number on the See your AI traffic tile.** Visits AI tools actually sent to your site in the last 28 days. This is the only figure on the page that counts people rather than answers, and it stays empty until Google Analytics is connected. ## When it goes wrong | You see | Why | What to do | |---|---|---| | A progress card instead of the numbers | Your brand is new. EOMA is reading your website, finding competitors and running the first scan. | Wait. The card fills in as each part finishes. | | "Brand analysis didn't finish" | EOMA could not read your website — usually a firewall or login wall, a page with too little text, or an AI provider being briefly unavailable. | Press **Try again**. If it fails twice, use **Edit my website URL**; a redirect or a `www` mismatch is the most common cause. Then contact support. | | "We couldn't load your dashboard" | The read failed. Your data is not gone; this screen could not fetch it. | Press **Try again**. If it repeats, contact support and quote the "Technical detail" line. | | A number reads "—" with *Fills with the first scan* | No scan has produced that figure yet. | Nothing to fix. It appears when the first scan finishes. | | "New actions arrive with the next scan" or "The next actions arrive on their schedule" | Nothing is waiting for you and EOMA is not working on anything: the board is clear. | Nothing to fix. Open [Actions](/docs/pages/actions) to see what is completed. | | "…of the 2 jobs EOMA does for you have been used" | On Free and Starter, EOMA does a set number of jobs for the life of the plan; they do not come back when a job finishes. On Free the other cards wait behind a lock. | On Starter, do any remaining actions yourself with the steps inside them; on either plan, upgrade so EOMA does more. | | "…EOMA runs 3 of your jobs at a time; all three are running now" | Every place your plan gives you is held by a job in flight. | Wait for one to finish; then you can start the next. | | The sentence at the top has not changed in weeks | The sentence follows the score, and the score moves slowly by design. | Read the arrow beside the score instead — it moves before the sentence does. | ## Related - [Actions](/docs/pages/actions) — the board this page sends you to - [AI provider visibility](/docs/pages/ai-provider-visibility) — the score, split into its two halves - [Your first week](/docs/start-here/your-first-week) — what to expect day by day --- # AI provider visibility > What ChatGPT, Claude, Gemini and Perplexity say about your brand — your score, who they name instead, and what to do about it this week. Source: https://eoma.ai/docs/pages/ai-provider-visibility Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - EOMA asks four AI engines a fixed set of questions about your market every week and keeps every answer. - The page reads top to bottom in two halves: everything above the orange line is measurement, everything below it is what to do. - The one number to watch is the **Overall AEO score**, and the two figures under it matter more than the score itself. ## Before you start - Your first collection must have finished. Until then the page says "Preparing your insights" and nothing else. Nothing is broken; the engines have not answered yet. - Your website address must be saved in your brand profile — it is how EOMA tells "AI cited you" from "AI cited a rival". See [Brand profile](/docs/settings/brand-profile). - Nothing else. This page is not restricted by plan. ## Steps ### 1 — Open it from the sidebar ``` ┌─ sidebar ───────────────────────────┐ │ Dashboard │ │ ▼ AI Visibility │ │ ● AI Provider Visibility ← here│ │ Your Brand on Google │ │ ▶ SEO │ └─────────────────────────────────────┘ ``` The parent row **AI Visibility** lands here too, so either click works. ### 2 — Read the score, then the line under it The score alone tells you little. The sentence under it — "knows you 84.2 · suggests you 0.0" — is the finding. ### 3 — Scroll to the orange line Above it is what was measured. Below it is what EOMA thinks you should do about it. With two minutes, read the score at the top and the three to-dos at the bottom. ## What you'll see ### The chapter line — "AI Visibility · Asking AI · ChatGPT · Claude · Gemini · Perplexity · Google AI Overview" An indigo stripe listing exactly who was asked, then a short explainer card you can close with the ×. ### "Overall AEO score" and "Responses analyzed" The two headline figures, top left, each with a small ⓘ for its definition. ### The switch — "Providers" / "Breakdown" Top right. It swaps what the narrow right-hand column shows, and nothing else. ### "AEO score over time", and the narrow column beside it The wide chart is one line per AI engine, week by week. The column beside it is either the four engines with a percentage each (**Providers**) or your score split across the eight kinds of question EOMA asks (**Breakdown**). ### "Share of Voice — Where the conversation lands" A donut: your slice highlighted, rivals with their own, small ones folded into "More competitors". The chips above it — Self, Competitors, Others, My tracked list — change what is counted, and the ⓘ in its heading **flips the card over** to "How high the bar actually is". ### "Where AI looks for answers" Beside the donut, the websites AI reads when it answers questions about your market, ranked. A different question from the donut: that one ranks who gets *named*, this ranks whose *link* the engine reaches for. ### The question list Full width, one row per question asked, showing the engines, the question, its category, your top rival on it, and how you did. Clicking a donut slice rewrites it to that rival's rows. ### The orange line — "What to do about it · Signals & Insights" The divider between measurement and advice. Under it, two cards: **Signals** ("What the AI answers are telling you this week"), drawn straight from the answers one finding at a time, and **Insights** ("Three things to do this week"). **The Insights card is not being written at the moment.** The weekly write-up behind it has not run since May 2026, so this card stays on its waiting message no matter how long you leave it. Nothing about your brand causes this and there is nothing to fix on your side. Your actual to-dos are on the [Actions](/docs/pages/actions) board, which is produced separately and is unaffected. Signals beside it is live. A band at the very bottom links to your website audit — see [Web presence](/docs/pages/web-presence). ## Reading the numbers **Overall AEO score — for example 25.3.** How visible you are across AI answers, from 0 to 100. It carries no `%` because it is not one: the larger half of it is scored against what the strongest brand in *your* market actually reaches, and measured leaders top out between 10 and 47. So 25.3 is not "a quarter of the way to perfect" — it is a position relative to real rivals. The method is in [How visibility is scored](/docs/concepts/how-visibility-is-scored). **"knows you 84.2 · suggests you 0.0" — the line under the score.** Worth more than the score above it. The first figure is how well AI answers when the question already contains your name; the second is how often AI brings you up when it does *not*. A brand at 84.2 / 0.0 is famous and never recommended — everyone who already knows you gets a good answer, and nobody discovers you. Closing that gap is what the actions at the bottom of the page are for. **Responses analyzed — for example 525, "8 runs · 28 days · 1,656 all-time".** The score was computed from 525 AI answers collected over eight sweeps in the last 28 days; "all-time" is every answer ever collected for you. The window never stretches to find data, which is why a quiet month reads as "not measured" rather than as an old number dressed as today's. **A percentage per engine in the Providers column — for example ChatGPT 31.4%.** Across every sweep ever run, ChatGPT mentioned your brand in 31.4% of its answers. This is a lifetime mention rate, not the score, so it is not expected to match the top of the page and a gap is not an error. Engines disagreeing with each other by a wide margin is itself a finding: the facts about you exist somewhere but are not reaching every model. **Your slice of the Share of Voice donut — for example 3.1%.** Of every brand name AI dropped into these answers, 3.1% of the mentions were yours. Flip the card with the ⓘ for the other half of that sentence: if the leader in your market reaches 58%, 3.1% is a long climb; if the leader reaches 9%, your market is wide open and 3.1% is better than it looks. **"#68 of 591" in Where AI looks for answers.** Of the 591 websites AI read to answer market questions, yours ranks 68th by how often it was used. Links to you from questions that already said your name are excluded on purpose — being cited when the question names you is handed to you, being cited when it does not is earned, and only the earned kind moves this ranking. **Outcome in the question list.** Each row is judged in the unit its own category measures, which is why the wording changes line to line: "4 affirmed" (answers that named you and backed it up), "1 unknown" (the model named you only to say it had no information about you — not a win), "#2 of 12" (where you placed on a shortlist and how long that list was), "cited 4×" (your site was the source, which for buying-guide questions *is* the result). Its Score is not sentiment: you can be described warmly and still lose the shortlist. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Preparing your insights" | Your first collection has not produced dashboard-ready answers yet. | Leave the page open or come back later. There is nothing to fix. | | "not scored yet" under the score | No scoring run has covered your brand. Usually a brand in its first days. | Wait for the first full sweep. If it persists past a week, contact support. | | "partial — not measured: sentiment" where the knows-you line should be | Some kinds of question produced nothing to score in this window. | The score is real but covers less ground than usual. Nothing to fix. | | "couldn't load — this is not the same as having no score" | The read failed. You may well have a score; this page could not fetch it. | Refresh. The wording is deliberate — do not read it as a zero. | | "The trend appears once you have a second scored window", or "Not enough comparable history yet" | One sweep cannot make a line, and sweeps that measured different kinds of question cannot honestly be joined. | Wait for the next collection. Nothing to do. | | "The run finished — AI's answers haven't named a single brand yet, not even a rival" | The donut is empty because the answers named nobody, not because collection is late. | Read the question list instead — it shows what the engines actually said. | | Your own site missing from "Where AI looks for answers" while other pages show it as cited | Every citation you have came from a question that already named you, which this card excludes. | Work the to-dos on your [Actions](/docs/pages/actions) board — those target the questions where a citation would count. | | "Your to-dos have not been written yet" in Insights | The weekly write-up behind this card has not run since May 2026. Waiting will not clear it. | Read the Signals card beside it, and take your to-dos from the [Actions](/docs/pages/actions) board instead. | ## Related - [How visibility is scored](/docs/concepts/how-visibility-is-scored) — what the eight question types are and how each is scored - [Share of Voice](/docs/pages/share-of-voice) — the same competitive question asked of Google rather than AI - [Actions](/docs/pages/actions) — where the three to-dos at the bottom become work EOMA can do --- # Your brand on Google > What Google actually shows when someone types your brand name — the AI answer, the sources behind it, and every result on the page. Source: https://eoma.ai/docs/pages/your-brand-on-google Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 Available on: growth, enterprise --- ## In 10 seconds - This is a real search for your brand name, run from your own market, kept exactly as Google returned it. - One badge under the answer tells you the thing that matters most: whether Google used your own website to describe you, or wrote about you from other people's pages. - It refreshes about once a week. The date beside the market tells you when Google actually answered — never when the page was loaded. ## Before you start - Your brand needs a **Growth plan or higher**. On Starter and on the free plan this page shows a lock. See [Plan and usage](/docs/settings/plan-and-usage). - Your website address must be saved in your brand profile. That address is what EOMA compares against Google's sources — without it, the "did Google use your site" badge has nothing to match. See [Brand profile](/docs/settings/brand-profile). - Your first capture must have run. It completes within 24 hours of setup, then repeats weekly. ## Steps ### 1 — Open it from the left sidebar ``` ┌─ sidebar ─────────────────────┐ │ ▾ AI Visibility │ │ AI Provider Visibility │ │ ● Your Brand on Google ← │ │ ▸ SEO │ └───────────────────────────────┘ ``` ### 2 — Check which market you are reading The pill at the top of the page names the country and language the search was run from. If you sell in more than one country, that pill becomes a dropdown — pick the market you want. ``` ┌──────────────────────────────┬───────────────────────────────┐ │ 📍 Turkiye · tr ← pick│ Google answered 12 Aug · 2d ago│ └──────────────────────────────┴───────────────────────────────┘ ``` Google answers the same brand name differently in different countries. One EOMA brand asked from Turkiye got a flights module and one result of its own; the same name asked from Germany got a full AI answer and two. The wrong market gives you a true answer to a question none of your customers asked. ### 3 — Read the badge under the answer Skip to the coloured line below Google's answer before you read anything else. Green means Google quoted your website. Amber means it described your business without opening your site once. That single line decides what you do next. ## What you'll see ### The market pill and "Google answered *date*" Side by side at the top. The date is the moment Google produced this answer, and the "2d ago" beside it counts from that moment. ### "*N* daily snapshots kept for this market" A small line under the pill, shown once you have more than one capture. EOMA keeps every past answer, so you can see how Google's description of you changed. ### "What Google's AI says about you" The largest block on the page. Your brand name appears as a chip on the right — that is the search that was run. Below it is Google's answer, word for word, with the small numbered circles Google uses to cite its sources. Each circle opens the page it came from. ### The badge, and "Other sites inside your own answer:" Directly under the answer: either *"Google used your own site to write this answer"* or *"Google wrote this about you without using your site once"*. Under that, the other websites Google read while writing about you. ### "Where it got this" The right-hand column. Every source Google used, as a card with the site name, the page title and a line of the text taken from it. Your own site is marked with a star and a green tint. Four are shown; "Show all *N*" opens the rest. ### "Your results page" A screenshot of the real page. Click it to expand. ### "Your brand card on Google" The box with your logo, description and links that Google shows beside search results for established brands. If you do not have one, this panel turns amber and says so. ### "What people ask about you" Google's own suggested questions on your name. If any of them are about whether you are trustworthy, the panel turns amber and counts them. ### "Who owns your name" Every result on a search for your brand name, in order. Yours are green, complaint and review sites are amber, everything else is plain. ### "Everything else Google showed" Chips naming the other blocks that appeared — maps, news, videos, images, forum discussions — with a count for each. ## Reading the numbers **"Google answered 12 Aug · 2d ago".** Two days ago is normal. This page refreshes weekly, so anything up to about seven days is the schedule working. Past that, see the last row of the table below. **"Google wrote this about you without using your site once".** Google described your business from other people's pages, so everything a customer reads about you was written by someone else. The fix is on your own website: clear answers to the questions in the panel below, on pages Google is allowed to read. **"3 of the top 20 results on your own name are yours".** Seventeen of the first twenty things a person sees when they search your name belong to somebody else — directories, marketplaces, review sites, competitors. Every extra page you own on your own name is one fewer chance for a customer to leave through their link. **"2 of these 8 are trust questions".** Two of the questions Google suggests about you are people checking whether you are legitimate. A page that answers that plainly, in your own words, is usually worth more than anything else you could write that month. **A source card count, for example "Show all 9".** Google read nine pages to describe you. If none of them are yours, those nine sites are shaping your reputation — and getting mentioned on any one of them moves you. ## When it goes wrong | You see | Why | What to do | |---|---|---| | A lock and "Upgrade to Growth" | Your brand is on Starter. This page belongs to Growth. | Open Settings → Plan and usage. The first capture runs within 24 hours of upgrading. | | "Your first capture is on its way" | No capture has completed yet. This is timing, not a restriction. | Come back within 24 hours. Nothing to fix. | | "Google's AI did not return an answer for this brand name in this market." | Google did not produce an AI answer for that search that day — common for smaller brands and smaller markets. | Read "Who owns your name" and the screenshot instead; both are still captured. | | "No screenshot for this capture." | The screenshot step failed for that run. The answer and the sources are unaffected. | Wait for the next weekly capture. | | "Google answered" shows a date over a week old | The weekly capture did not run for your brand. | Contact support with your brand name and the date shown. | | Your brand card panel is amber | Google holds no brand card for your name. | This is one of the highest-value things on the page to fix. Wikidata, consistent business details, and linked social profiles are what build it. See [Web presence](/docs/pages/web-presence). | ## Related - [Web presence](/docs/pages/web-presence) — the checks that decide whether Google and AI engines can read your site at all - [AI provider visibility](/docs/pages/ai-provider-visibility) — the same question across ChatGPT, Claude, Gemini and Perplexity --- # Keyword portfolio > The words people type into AI engines around your category, how much demand each one carries, and which ones are rising or falling. Source: https://eoma.ai/docs/pages/keyword-portfolio Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Each row is one search term, with its estimated monthly volume on AI engines and a twelve-month trend line. - Terms are labelled **Brand**, **Rival** or **Category**, so you can see at a glance how much of the demand is your name and how much is the market's. - This list is collected once, when your brand is set up. It does not refresh weekly — check the "Updated" date before you treat a number as current. ## Before you start - Nothing to switch on. The list is collected during setup, alongside your first AI visibility run. - Your competitor list decides which rows get the **Rival** label. If it is empty, rival terms are labelled Category instead. See [Competitors](/docs/pages/competitors). ## Steps ### 1 — Open Share of Voice and use the link at the bottom This page has **no entry in the left sidebar**. The reliable way in is through SEO → Share of Voice, then the link under the keyword tables. ``` ┌─ sidebar ──────┐ ┌─ Share of Voice ────────────────────────┐ │ ▸ AI Visibility│ │ … │ │ ▾ SEO │ │ View the full AI keyword portfolio → │ │ ● Share of │ │ ← click│ │ Voice ← │ └─────────────────────────────────────────┘ └────────────────┘ ``` Typing `/aeo/keywords` after your EOMA web address works too, and is worth a bookmark. ### 2 — Read the two numbers at the top How many terms are being tracked, and how much monthly demand they carry between them. The sentence beside them tells you how many are rising and how many are falling. ### 3 — Open a row to see its year Click any row. It expands into a larger chart and twelve month-by-month tiles, so you can tell a real trend from one unusual month. ## What you'll see ### "Keyword portfolio" and the headline The eyebrow label, then your brand name followed by "AI search volume and 12-month trend". The information icon beside the eyebrow explains the source. ### "Updated · *date*" Top right, beside a green dot. The day this list was collected. Read the third row of "When it goes wrong" before treating it as recent. ### "Keywords" and "Monthly AI searches" Two figures side by side, with a sentence to their right naming how many terms are rising and falling. ### "Trend Comparison" A single chart holding the six highest-volume terms over the months EOMA has data for. The legend on the left is clickable: click a term to hide its line, hover to bring it forward. Hovering the chart shows every value for that month at once. ### "AI search volume list" The table. Four columns — **Keyword**, **12-month trend**, **Volume**, **Δ %** — and a coloured tag on every term reading Brand, Rival or Category. Brand rows are tinted. Clicking a row opens its full year underneath. ## Reading the numbers **Keywords — for example 50.** Fifty terms are being tracked around your brand and your category. The page loads at most 50, so a count of exactly 50 means "50 or more", not "that is the whole market". **Monthly AI searches — for example 74,000.** Those fifty terms are typed into AI engines around 74,000 times a month between them. That is the size of the conversation you are competing inside — not traffic you would receive. Compare it with your own brand rows: if your name carries 400 of the 74,000, then 99% of the demand in your category is being asked without you in mind. **A Brand tag.** The term contains your brand name — someone already looking for you. **Rival** means it contains a competitor's name; that demand exists and is being answered by somebody, and it is not you. **Category** is the generic demand — "villa rental Antalya" rather than any company name. Category rows are where new customers come from; Brand rows are where existing ones are kept. **Δ % — for example +38%.** The change from the second-to-last month to the last month in that term's series. Anything between −5% and +5% is shown grey and treated as flat, because month-to-month noise at that size means nothing. A "—" means the term has no measurable volume at all. **The sparkline in "12-month trend".** The same twelve months drawn small. Green means the last step was up, red down. Use it to spot a term whose Δ % looks calm but whose year has collapsed — the single-month number cannot show that, and the line can. **"Top 6 keywords by volume — 12-month trend".** The comparison chart shows only the six largest terms, and only those with at least two months of data. A term missing from the chart is not missing from the table — it is only absent from the top six. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No keyword portfolio yet." | The keyword collection has not run for your brand. It runs during setup. | If your brand finished setup more than a day ago, contact support. | | "Couldn't load data. Refresh to try again." | The read failed. It does not mean you have no keywords. | Refresh. If it repeats, contact support. | | "Updated ·" shows a date weeks or months old | This list is collected once during setup and has no recurring refresh. Every brand in EOMA today holds exactly one collection date, and the oldest is over three months back. | Treat the volumes as a baseline for planning, not a weekly tracker. For a number that moves every week, use [Share of Voice](/docs/pages/share-of-voice). | | A competitor's term tagged Category, not Rival | That competitor is not in your competitor list. | Add them on [Competitors](/docs/pages/competitors); the tag follows the list. | | No "Trend Comparison" chart | Fewer than two of your terms have at least two months of history. | Nothing to fix. The chart appears once the data supports it. | | Δ % reads "—" on a row with volume | The term's monthly series is missing or has a single month. | Open the row: if the twelve tiles are empty, that term has no history yet. | ## Related - [Share of Voice](/docs/pages/share-of-voice) — where this page is linked from, and the weekly view of the same market - [Portfolio](/docs/pages/portfolio) — the Google-side view of the keywords you rank for --- # Web presence > A scan of your website from an AI crawler's point of view — what it can read, what it cannot, and what is missing. Weekly, plus a re-check button you can press once a day. Source: https://eoma.ai/docs/pages/web-presence Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-29 --- ## In 10 seconds - Around thirty checks run against your site every week, grouped into what AI needs to *reach* you and what it needs to *trust* you. **Fixed it? Check again**, top right, re-runs all of them on demand — once a day, about a minute each time. - Each check reads Complete, Partial, Missing or one of four other states, with a count above each group. A check you have already fixed since the last scan is marked **Not re-measured** until you re-run it. - The badge at the top is decided entirely by the foundational group. One missing item there is enough to keep it off Healthy. ## Before you start - Nothing to switch on. The first scan runs during setup, then repeats weekly on every paid plan. - On the free plan the weekly repeat does not run: the re-check button is how you refresh, and it can be pressed once every 30 days (once a day on a paid plan). - Your website address must be saved in your brand profile — that is the site the scanner visits. See [Brand profile](/docs/settings/brand-profile). - Most of what this page reports is changed on your own website, not inside EOMA. See [What EOMA does and what you do](/docs/start-here/what-eoma-does-and-what-you-do). ## Steps ### 1 — Open AI Provider Visibility and find the audit strip This page has no sidebar entry of its own. It is reached from the card near the bottom of the AI Provider Visibility screen. ``` ┌─ AI Provider Visibility ─────────────────────────────┐ │ … │ │ WEB PRESENCE AUDIT │ │ See your web presence technical checklist │ │ 21 of 32 checks complete · 6 partial · 5 missing │ │ [Needs attention] │ │ View audit → ← click │ └──────────────────────────────────────────────────────┘ ``` ### 2 — Read the badge, then the group counts The badge is a verdict on the foundational group only. The three counts above each column tell you where the work is. ### 3 — Start with anything Missing in the first column Work left to right. A missing item in "Foundational technical" can stop an AI engine reading your site at all, which makes everything in the other columns unreadable too. ## What you'll see ### "Web presence audit" and "Web presence signals" The eyebrow label with an information icon, then the heading. The icon explains what the scan covers. ### The health badge and "Last checked *date and time*" Top right. The badge reads **Healthy**, **Needs attention**, **Critical** or **Scan failed**. Beside it, the exact moment the scan ran. ### "Checked *N* days ago" and the re-check button Page header, top right. How old this audit is, in days, and the button that retakes it. One press per day; when you have used it, the line beneath names the hour your next one opens. The scan takes about a minute and the page updates on its own — you do not have to reload. ### "Your work is not in these numbers yet" A banner above the checks, and it only appears when it has rows to point at. It means you closed an EOMA action for one of the checks below *after* this scan was taken, so those rows describe your site as it was before your fix. They carry a **Not re-measured** chip. The chip does not claim the check now passes — we do not know that until the site is read again. ### "Foundational technical" The first column — the checks that decide whether an AI crawler can reach and parse your site at all: `llms.txt`, `robots.txt`, AI crawler access, `sitemap.xml`, Organization schema, sameAs links, Wikipedia, Wikidata, canonical URL, OpenGraph tags, and three graded checks on your schema and llms.txt quality. ### "Authority + entity" The second column — the checks that decide whether AI treats you as a credible, specific business: FAQ, Article, Breadcrumb and LocalBusiness schema, `mainEntityOfPage`, HTTPS, homepage response time, JS render requirement, hreflang, favicon, RSS/Atom feed, security headers, and the graded content checks (E-E-A-T signals, BLUF answer quality, topic cluster coverage, differentiator visibility, page intent taxonomy, conversational query coverage, content freshness). ### The count line above each column "18 complete · 6 partial · 5 missing", with "· 3 other" added when checks came back N/A or errored. ### Each check row The check name on the left, its state on the right with a matching icon. States are **Complete**, **Partial**, **Minimal**, **Outdated**, **Missing**, **Blocked**, **Error** and **N/A**. ### "Scanned: *your site* · *N* pages crawled" The footer. The exact address that was visited, and how many pages were read. ## Reading the numbers **"21 of 32 checks complete · 6 partial · 5 missing".** Eleven of the things AI engines look for on a website are absent or half-done on yours. Partial usually costs less to fix than Missing — the thing exists and needs finishing. **The badge, "Needs attention".** One or more foundational checks are not Complete. That is the whole rule: **Healthy** requires every foundational check to be Complete or not applicable, and **Critical** appears once three or more of the essential ones are missing or blocked outright. A badge stuck on Needs attention with only one missing item is not the same problem as one with eight — read the column, not the badge. **"AI crawler access — Missing".** Your `robots.txt` is turning away the crawlers that ChatGPT, Claude and Perplexity use. Nothing else on this page matters until that is fixed: an engine that cannot fetch your pages will describe you from other people's sites instead. This is the single highest-cost failure in the list. **"Organization schema — Missing".** Your site never states, in a form a machine reads, what your business is called, where it is, and how to contact it. AI then guesses from your page text, and guesses inconsistently. **"Wikidata — Missing".** No structured entity exists for your brand. Wikidata feeds Google's Knowledge Graph, which is what produces the brand card on [Your brand on Google](/docs/pages/your-brand-on-google). Missing here and a missing brand card there are usually the same problem. **"llms.txt — Missing".** A short file that tells AI engines which of your pages matter and how to describe you. It is new, cheap to add, and almost nobody in most categories has one. **"· 45 pages crawled".** The scanner read 45 pages of your site to reach these answers. A low number on a large site can mean pages are not linked from anywhere the crawler could follow — which means AI is not finding them either. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No web presence audit yet — one will appear after the next scan." | Your brand has not been scanned. The first scan runs during setup. | If setup finished more than a week ago, contact support. | | "Couldn't load the web presence audit. Refresh to try again." | The read failed. It does not mean the scan failed. | Refresh. If it repeats, contact support. | | The badge reads "Scan failed" and there are no check rows | The scanner could not read your site at all — usually an unreachable address, a redirect loop, or a firewall blocking automated visitors. | Open the address in the footer in a private browser window. If it loads for you but not for the scanner, your host or security service is blocking it. | | Only two columns, no "Content quality" group | The audit defines a third group, but no check is currently filed under it, so it does not render. The content-quality checks themselves still run — they appear inside the two columns you can see. | Nothing to fix, and nothing is missing from your results. | | The count in the entry card says 26, the page shows more | The tooltip text predates six checks added in July. | Trust the column counts on this page. | | A check reads "Error" | That one check could not complete — often a timeout on a slow page. | Press **Fixed it? Check again**. A single Error does not change the badge, so the weekly scan will also clear it on its own. | | "One re-check a day" under the button | You have already re-checked in the last 24 hours. | The line names the exact time your next one opens. The weekly scan still runs regardless. | | A row is marked **Not re-measured** | You completed an EOMA action for that check after this scan ran, so the state beside it describes your site *before* your fix. | Press **Fixed it? Check again**. Until you do, treat that row as unknown rather than as a finding. | ## Related - [Your brand on Google](/docs/pages/your-brand-on-google) — what these signals produce, or fail to produce, on a real search - [AI provider visibility](/docs/pages/ai-provider-visibility) — the screen this page is reached from - [Actions](/docs/pages/actions) — where EOMA turns a missing check into work it can do for you --- # SEO summary > One screen that answers where your brand stands on Google — market share, ranking keywords, clicks earned, and the next step worth taking. Source: https://eoma.ai/docs/pages/seo-summary Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - This is the front door to the five SEO screens — every card carries a link into the detail behind it. - It adds no new numbers of its own. Each figure is one of the detail screens, shortened. - A card with nothing in it says why it is empty and when it will fill, instead of showing a zero. ## Before you start - Keywords must already exist for your brand. EOMA builds that list for you during setup, by hand — it is not started by your signup. Until it is done, this screen says your portfolio is being built. See [Portfolio](/docs/pages/portfolio). - Your first Google ranking scan must have run. It completes within 24 hours of the keyword list being ready, then repeats weekly. - Nothing else. Every plan sees this screen, and there is no upgrade lock on it. Your plan changes how many keywords are measured, not whether the screen works. See [Plans and limits](/docs/start-here/plans-and-limits). ## Steps ### 1 — Open SEO in the left sidebar The section header itself lands here. You do not need to pick a sub-item. ``` ┌─ sidebar ──────────────┐ │ AI Visibility │ │ ● SEO ← click │ SEO summary │ │ Share of Voice │ │ Opportunities │ │ Portfolio │ │ Rank Tracking │ │ Competitors │ └────────────────────────┘ ``` ### 2 — Read the sentence at the top One line, in plain numbers: how many keywords you appear on, what they are worth in clicks, where you sit against the field. If you read nothing else, read that. ### 3 — Follow one card into its detail screen Each card ends with a link — *Rank tracking →*, *Visibility →*, *Opportunities →*, *Portfolio →*. The card is the summary; the link is the working screen. ## What you'll see ### The verdict line A full-width sentence under the title. It names the keywords you appear on, the clicks those positions are worth, your position in the field, and how many keywords sit one step below page 1. On a brand with no rankings yet it says so instead, and the rest of the screen still renders. ### The four numbers — "Your market share", "Ranking keywords", "Est. clicks/month", "Winnable" A row of four, left to right. Each carries a smaller line under it saying what it was measured against. ### "Position distribution" · "How is the share split across the keywords you compete on?" · "Since the last scan" The first card row. A colour bar of where your positions sit, a ladder of who takes the clicks on the searches you share, and a movement count. ### "Closest to page 1" · "Demand and intent" The second card row. Five keywords ranked 4–20, page 2 first; then how much of your own ceiling you take, and what kind of demand you are chasing. ### "Next step" One line at the bottom with a coral bar beside it. Always exactly one — chosen by a fixed order of what usually blocks a brand, never a list to pick from. ## Reading the numbers **Your market share — for example %11.9, "3 of 14 domains".** Of the clicks available on searches where a rival also shows up, you take about twelve in every hundred, and thirteen other websites are splitting the rest. It is deliberately *not* measured on searches you appear on alone — a search nobody else competes for is not a market, and counting it would push this number toward 100%. The smaller line tells you how many keywords it rests on. **"<0.1%" instead of a number.** You hold a real but tiny share. This is a different fact from `—`, which means there is no contested keyword to measure at all, and the screen keeps them apart on purpose. **Ranking keywords — for example 37/50.** You appear somewhere in Google's top 20 on 37 of the 50 keywords being measured every week. The other 13 are being checked and are not showing up. **Est. clicks/month — for example 214, "%18 of the ceiling".** What your current positions are worth per month, using a fixed click-rate curve by position. The ceiling is what you would take at position 1 on everything measured — so %18 says there is roughly five times your current traffic still sitting in keywords you already track. Where the ceiling cannot be worked out, it says so rather than printing a percentage of nothing. **Winnable — for example 12, "%24 of monitored".** Keywords where you rank between 4 and 20. Close enough that a better page can move them, far enough that they earn little today. This is the pool the "Next step" line usually points at. **"Position distribution" — Top 3 · Page 1 (4–10) · Page 2 (11–20) · Past 20 · Not ranking.** Counts, not percentages. *Past 20* is shown in a warning colour and is not a weaker version of page 2 — a position at #110 is a different problem from a #12. **The ladder, "you" marked in coral.** Each bar is one domain's share of the clicks on the keywords you share with rivals. Yours is always on the list even when it sits below the five shown. **"Since the last scan" — Improved · Declined · New entry.** Each keyword is compared against its own two most recent measurements, not against a calendar week. The line underneath names how many keywords qualified; everything else was measured for the first time. **"Closest to page 1" — for example `antalya villa kiralama · #14 · 6,600`.** Position 14, six thousand six hundred searches a month. Page-2 keywords are listed before page-1 ones on purpose: a #14 is one push from the fold, a #6 is already earning. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No keywords for *your brand* yet. Your portfolio is built during setup and appears here when it is ready." | Nobody has built the keyword list for your brand yet. This step is run by hand at EOMA and is not triggered by signing up, so a new brand can sit here through no fault of yours. | Ask support to start it. No SEO screen fills until it is done. | | "Loading portfolio…" or "Could not load the keyword portfolio." on this screen | The read that gates this page failed. The wording says *portfolio* even though you are on the summary — that is our copy being wrong, not you being on the wrong screen. | Refresh. If it repeats, contact support. | | "This is measured on searches where **your rivals also appear**. You have none of those yet" | Your rankings are mapped, but no competitor page has been captured on any of them. | Nothing to fix. It fills as weekly scans accrue. | | "Movement needs **two measurements** of the same keyword. None qualify yet." | Only one scan has covered your keywords so far. | The first comparison lands on the next scan. | | Market share shows `—` | No keyword of yours has a rival measured on it, so there is no honest share to print. | See [Share of voice](/docs/pages/share-of-voice) for what is measured instead. | | "ceiling not measurable yet" under Est. clicks/month | Every keyword being measured has no reported search volume. | Add keywords that carry volume from the Portfolio screen. | ## Related - [Share of voice](/docs/pages/share-of-voice) — the full version of the market-share number - [Opportunities](/docs/pages/opportunities) — where the "Winnable" keywords are worked - [Portfolio](/docs/pages/portfolio) — the whole keyword list, and where you change it --- # Share of voice > How much of the clickable Google demand you hold on the searches you actually share with rivals — and which websites hold the rest. Source: https://eoma.ai/docs/pages/share-of-voice Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Share is measured only on keywords where a rival was actually captured — a search nobody else competes for is not a market. - The rivals listed here are the ones Google shows us. They are not the ones you chose to follow; that list lives on Competitors. - The bottom-right link is one of the only ways into your AI keyword portfolio. ## Before you start - Your keyword list must exist and your first Google scan must have run. Both are set up for you — see [Portfolio](/docs/pages/portfolio). - Competitor pages must have been captured on at least one of your keywords. Without that there is a coverage picture but no share, and the screen says so rather than inventing one. - No plan requirement. Every plan opens this screen; your plan sets how many keywords get a full search-results page fetched each week. See [Plans and limits](/docs/start-here/plans-and-limits). ## Steps ### 1 — Open SEO → Share of Voice ``` ┌─ sidebar ──────────────┐ │ ● SEO │ │ SEO summary │ │ Share of Voice ← click │ Opportunities │ └────────────────────────┘ ``` ### 2 — Read the big percentage together with the line under it The line under it names how many keywords the percentage was measured on. The two belong together — quoting the percentage without the count is the fastest way to mislead yourself. ### 3 — Work down to the two tables at the bottom *Where your visibility comes from* is what to defend. *Where you're losing — biggest gaps* is what to attack, and the column on its right names who is currently taking it. ## What you'll see ### "Search engine visibility" and the big "Share of Voice" number The title strip, then the headline percentage set large, sometimes with a green or red pill beside it reading *N pts vs last month*. Under it: how many keywords it was measured on, and a sentence putting the same figure in words. ### The five-column rail — "Measured", "Ranking", "Top 3", "Striking (4–10)", "Avg position" A row of five figures across the width of the screen, describing coverage rather than share. ### "Share of voice · where rivals are visible" and "Who shares your keywords" Side by side. A ring chart with your slice in mint and a legend naming every slice, then a list of the rival domains with their share and average position. Under the list, a line pointing at the Competitors screen. ### "Where your visibility comes from" and "Where you're losing — biggest gaps" Two tables at the bottom, six rows each, with the same columns: Keyword, Google, AI, You, and either Top rival or Who leads. ### The footnote and "View the full AI keyword portfolio →" Bottom left explains the AI column; bottom right is a link into your AI keyword portfolio at `/aeo/keywords`. That link matters — the AI keyword screen has no sidebar entry of its own, so this is one of the few doors into it. See [Keyword portfolio](/docs/pages/keyword-portfolio). ## Reading the numbers **Share of Voice — for example 10.1%, "of the 91 keywords where we can see rivals".** On those 91 searches, about one click in ten comes to you. The count is the honest part: if you are measured on 518 keywords but only 91 have a rival captured, this percentage describes those 91 and nothing else. A brand measured on its own uncontested keywords would read near 100% and mean nothing. **"▼ 7.2 pts vs last month".** The change is worked out on the keywords present in *both* months, then both months are recalculated on that shared set. Without that, adding or removing keywords would move the line exactly like a real loss. If fewer than five keywords are common to both months, no pill appears at all — a missing number is honest, a wrong one is not. **"Measured" — for example 518.** Keywords we hold a Google position for. This is **not** the number your plan pays for: that one is on [Rank tracking](/docs/pages/rank-tracking), and it is usually far smaller (25 to 100 depending on plan). Three different keyword counts exist across these screens and this one is the widest. **"Ranking" — for example 314 (61%), and "Top 3" — 27.** Of the 518 measured, you appear in the top 20 on 314, and hold the top three on 27. **"Striking (4–10)" — for example 46.** Forty-six keywords sit on page 1 but below the top three. These move with a better page rather than a new one. Note that the Opportunities screen uses the same word for a wider band, 4–20 — the two screens count different things under one label. **"Avg position" — for example #8.4.** Averaged across the keywords you rank on only, so it cannot be read as "how visible am I" on its own — pair it with Ranking. **The ring chart slices.** Your slice, up to five rival domains, and a grey *others* remainder — all shares of the same contested demand, so they add to 100. **A rival row — for example `sahibinden.com · 79 kw · 16.6% · #6.2`.** That domain shows up on 79 of your keywords, takes about seventeen clicks in every hundred, and averages position 6.2 where it appears. A domain must appear on at least **two** of your keywords to be listed — one shared keyword is a coincidence, and listing it made an unknown site outrank a real competitor. **▼ and ▲ beside a rival.** ▼ green means that rival is losing ground to you. ▲ red means they are gaining on you. The arrow describes their trend, not yours. **Missing platforms in the rivals list.** Instagram, YouTube, Wikipedia, job boards and forums are counted in the denominator — they really do take those clicks — but are kept out of the rivals list, because a site people search to find a job is not competing with you for a customer. **"6,600" in the Google column, a teal number in the AI column.** Monthly Google searches, then monthly answer-engine demand for the same keyword. The AI figure is context only; the share above is weighted by Google volume. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Your rankings are on the way" with a keyword count | Keywords are selected and the first Google scan has not run. It runs within 24 hours, then weekly. | Wait. If it still says this after two days, contact support — this screen cannot see when the last scan ran, so it cannot tell "not yet" from "ran and found nothing". | | "No Google rankings for *your brand* yet." | Nothing is in the monitored set at all, usually because the keyword list has not been built for your brand yet. That step is run by hand at EOMA. | Ask support to start it. | | The big number is `—` and the text reads "haven't captured competitor SERPs yet" | Your positions are mapped but no competitor page was captured, so there is no share to divide. | Nothing to fix. The coverage rail and both tables are still real. | | No "vs last month" pill | Fewer than five keywords are present in both months, or there is only one month of data. | It appears once two months share enough keywords. | | "Could not load SEO visibility." | The read failed. | Refresh; if it repeats, contact support. | | The rivals list shows six names while Competitors shows 0 | These are two different lists. This one is measured from Google automatically; Competitors is the set you chose to follow. | Track the ones that matter from [Competitors](/docs/pages/competitors). | | "Measured" here disagrees with "Monitored" on Rank tracking | They count different things: rank data held versus keywords your plan fetches a full results page for weekly. | Both are correct. Use Rank tracking for what you pay for. | ## Related - [Opportunities](/docs/pages/opportunities) — the keywords closest to a win - [Competitors](/docs/pages/competitors) — choose which rivals get followed week to week - [Keyword portfolio](/docs/pages/keyword-portfolio) — the AI-side keyword screen this page links out to --- # Opportunities > The Google keywords you can win next — ranked 4 to 20 today, plus the biggest gaps in your pool where you do not appear at all. Source: https://eoma.ai/docs/pages/opportunities Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - The first table is the short list: you already rank 4–20 there, so a better page moves you, not a new one. - The second table is whitespace — real gaps in your own pool, worth checking for fit before you write. - A position with a dotted underline was measured weeks ago and has not been found since. ## Before you start - Your keyword list and first Google scan must be in place. Both are set up for you — see [Portfolio](/docs/pages/portfolio). - Nothing else. Every plan opens this screen. Your plan sets how many keywords get a full search-results page fetched each week, which is what fills the *Who leads* column. See [Plans and limits](/docs/start-here/plans-and-limits). ## Steps ### 1 — Open SEO → Opportunities ``` ┌─ sidebar ──────────────┐ │ ● SEO │ │ Share of Voice │ │ Opportunities ← click │ Portfolio │ └────────────────────────┘ ``` ### 2 — Work the first table from the top The rows are already ordered by opportunity score. The top row is the recommendation; the priority bar on the right shows how far each row sits below it. ### 3 — Check the second table for fit before writing Whitespace rows are ranked by demand and winnability, not by whether they suit your business. Some are genuine gaps, some are off-theme. You are the one who can tell them apart. ## What you'll see ### "Search opportunities" and the big "Striking distance" number The title strip, then a large count with *keywords one push from page 1* beside it, and a sentence naming both that count and the whitespace count. ### The four-column rail — "Measured", "Holding top-3", "Striking (4–20)", "Whitespace" Under the headline, describing the whole keyword set rather than the rows shown. ### "Quick wins — striking distance" The main card. A table of up to twelve rows: Keyword, Google, AI, You, Who leads, Priority. Under it, one or two lines owning up to what the table cannot tell you. ### "Whitespace — the biggest gaps in your pool" The second card, up to eight rows, same columns. Its opening line changes with your data: where the top row has real volume it reads as demand, and where the pool is thin it says so and quotes its own largest number. ### The footnote and "← Back to share of voice" Bottom left explains the score; bottom right returns you to [Share of voice](/docs/pages/share-of-voice). ## Reading the numbers **Striking distance — for example 405.** Four hundred and five keywords where you already rank between 4 and 20. You have proven Google finds you relevant there, so the work is improving a page rather than earning a position from nothing. Only the top twelve are listed; the count speaks for all of them. **"Measured" — for example 518.** Keywords we hold rank data for. Not the number your plan tracks weekly — that one is on [Rank tracking](/docs/pages/rank-tracking) and is smaller. **"Holding top-3" — for example 27.** Already won. They are counted here and deliberately not listed, because a position you hold is not an opportunity, it is something to defend. **"Whitespace" — for example 96.** Keywords where you are not in the top 20 at all. Ranked by demand multiplied by winnability, so the top of this list is the biggest gap *inside your own pool* — which on a small pool can still be a small number. The card says which of the two you are looking at. **"of which N unconfirmed" under Striking.** Those positions were last seen more than three weeks ago. The weekly check has looked since and not found them, so treat them as history rather than as today. **A dash in the Google column instead of 0.** Google reported no measurable volume, which is not the same as no demand — roughly nine in ten keywords fall under every volume tool's floor. A zero would be a small lie in a column about demand. **The teal number in the AI column.** Monthly answer-engine demand for that keyword. Cross-channel context; it does not enter the score. **A position with a dotted underline — for example `#14`.** Your rank, last measured more than 21 days ago. The weekly check only returns keywords your domain holds in Google's top 100, so three misses in a row means three checks looked and did not find you. The row stays and stays sorted — an old measurement is still the best evidence we have — but it is marked so you do not act on it as current. **Under the keyword: `learn · KD 34 · /villa-kiralama`.** The search intent, the keyword difficulty (0–100, lower is easier) and the page of yours that actually ranks. That last part is the useful one: *your homepage ranks 14* is an instruction to build the page, where *you rank 14* is only a number. **"not checked" in the Who leads column.** We hold your position but never captured the full results page, so we cannot say who is above you. That is different from a dash, which means we looked and no relevant competitor was there. Pinning a keyword on [Portfolio](/docs/pages/portfolio) brings it into the weekly full-page sweep. **The Priority bar and its number, 0–100.** How this row scores against the single strongest opportunity on the screen — 100 is the best row, 40 is a bit under half as attractive. The score is demand (compressed, so a 100,000 keyword is not a hundred times a 1,000 one) × winnability × intent, with buying intent weighted above informational. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Your rankings are on the way" with a keyword count | Keywords are selected, the first Google scan has not run. It completes within 24 hours, then weekly. | Wait. If it persists past two days, contact support — this screen cannot see when the last scan ran. | | "No ranking data for *your brand* yet." | Nothing is in the monitored set, usually because the keyword list has not been built for your brand. That step is run by hand at EOMA and is not started by signing up. | Ask support to start it. | | "No striking-distance keywords yet" | You rank either in the top 3 or nowhere on everything measured. | Work the whitespace table instead, and add keywords from Portfolio. | | "No whitespace gaps surfaced" | You rank somewhere on every measured keyword that has competitor data. | Grow the pool from Portfolio. | | The quick-wins card says Google reports no measurable volume for these | None of the twelve rows shown has measured demand. | The rankings are still real. Treat them as early signal and add keywords with volume. | | Most of the Who leads column reads "not checked" | Those keywords are outside the weekly full-page sweep, which covers only your plan's tracked set. | Pin the ones you care about on Portfolio to bring them in. | | "Could not load SEO opportunities." | The read failed. | Refresh; if it repeats, contact support. | | Striking here is 4–20, but Share of voice says "Striking (4–10)" | The two screens use one word for two ranges. | Trust the range printed in the label you are looking at. | ## Related - [Portfolio](/docs/pages/portfolio) — pin a keyword to bring it into the weekly full-page sweep - [Rank tracking](/docs/pages/rank-tracking) — what your plan measures every week - [Share of voice](/docs/pages/share-of-voice) — how much of the market these positions are worth --- # Portfolio > Every keyword EOMA holds for your brand, what the list is worth in clicks, and the controls to add, pin or set aside keywords yourself. Source: https://eoma.ai/docs/pages/portfolio Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - This is the only screen where you can add a keyword, pin one, or set one aside. - Pinning spends from the same weekly budget the machine uses — what you pin always wins, one of its picks drops out. - The table opens filtered to the monitored set, not the whole pool. The filter chip is highlighted when it is narrowing. ## Before you start - Your keyword list must exist. EOMA builds it for your brand by hand during setup; it is not started by your signup, so a new brand can sit empty here for a while through no fault of its own. - Nothing else. Every plan opens this screen. Your plan sets two caps: how many keywords the pool may hold, and how many get a full Google results page fetched each week — the second is also the cap on how many you may pin. See [Plan and usage](/docs/settings/plan-and-usage). ## Steps ### 1 — Open SEO → Portfolio ``` ┌─ sidebar ──────────────┐ │ ● SEO │ │ Opportunities │ │ Portfolio ← click │ Rank Tracking │ └────────────────────────┘ ``` ### 2 — Add a keyword you want measured Type it into the box above the table and press Add. Anything is accepted, including terms with no reported search volume — those never come up in automatic discovery and are often the ones that convert. ``` ┌──────────────────────────────────────────────┬───────┬──────────────┐ │ Add a keyword you want tracked… │ Add │ 12 of 50 pinned│ └──────────────────────────────────────────────┴───────┴──────────────┘ ``` ### 3 — Pin what matters, set aside what does not Every row has two buttons on the right. **Pin** forces a keyword into the weekly measurement whatever the score says. **Remove** means "not my market" — it hides the row and stops the keyword coming back through discovery. Nothing is deleted: the *Set aside* filter lists everything you removed, with a **Put back** button. ## What you'll see ### "Demand · N monitored keywords" and the big clicks figure The top panel. One large number with *clicks/mo — what this portfolio is worth* beside it: the monthly clicks you would take at position 1 on every keyword being measured. ### "How much of the demand you cover" and "How much of the ceiling you take" Two meters under it. The first compares the searches you measure against all the searches in your pool. The second compares today's clicks against that ceiling. ### "Is this demand the buying kind?" A colour bar splitting your measured demand by intent, with the number of keywords the percentages rest on. It is hidden entirely when no intent is known. ### "Measured on the next scan" A dashed coral box, shown only when you have pinned something the weekly pass has not acted on yet. It sits directly under the add box, so a keyword you just requested is the first thing you see rather than something to hunt for in the table. ### The five-column rail — "Keywords", "Google demand", "AI demand", "Ranking", "Top 3" Counts for the whole pool, under the demand panel. ### The filters and the table A search box, an intent list, a status list (*All*, *Ranking*, *Not ranking*, *Monitored*, *Pinned by you*, *Set aside*), a **Closest to page 1** button, and the count of rows shown. Then the table: Keyword, Google, AI, KD, Intent, You, Yours. Your pinned rows are grouped at the top under *What you pinned*. ## Reading the numbers **The big clicks figure — for example 488.** If you held position 1 on every keyword currently being measured, those searches would send you about 488 visits a month. It is a ceiling, not a forecast — but it tells you whether this list is worth a quarter of work or an afternoon. **"How much of the demand you cover" — for example %64.** Of all the searches in your pool, 64% sit on keywords that are actually being measured each week. The line under it names how many pool keywords have no measurable volume at all; those can never be covered by this meter. **"How much of the ceiling you take" — for example %3.** You earn about 15 of a 488-click ceiling. A low number here is not a fault in the data; it is the gap between where you rank and position 1, which is exactly the work [Opportunities](/docs/pages/opportunities) lists. **The intent bar — for example `buy 12,400 · compare 3,100 · learn 900`.** Monthly searches by what the searcher wants. A portfolio that is mostly *learn* will grow traffic and not sales. If purchase-intent demand is exactly zero, the panel says so in words rather than leaving you to read a missing colour. **"Keywords 1,712 · 50 monitored".** Your pool holds 1,712 keywords; 50 get a full Google results page fetched every week. The pool cap and the weekly cap are different limits and both come from your plan — 200 to 10,000 for the pool, 25 to 100 for the weekly set. **"Google demand 214K", "AI demand 18.2K", "Ranking %61", "Top 3 27".** Pool-wide totals — monthly Google searches, monthly answer-engine demand, how much of the pool you appear on, and how much of that sits at the very top. They cover the whole pool, so they are larger than anything the meters above use. **A dash in the Google column.** Volume could not be measured. About nine in ten keywords sit under every tool's reporting floor, so this is common and does not mean nobody searches for it. **KD** beside it is keyword difficulty, 0 to 100, lower is easier. **The You column — `#14` with a short bar, or "not monitored".** The bar is scaled by how much traffic that position is actually worth, so it empties past #20 on its own. *not monitored* means the keyword was never measured, which is different from a dash, which means it was measured and did not appear in the top 20. **"12 of 50 pinned".** You have used 12 of your plan's weekly measurement slots on your own choices. When the budget is full, what you add still wins — one of the machine's picks drops out instead. Your own pins are never displaced. **Two badges on one row, "pinned" and "monitored".** They are independent facts and can disagree for up to a week: *pinned* is your decision, taking effect immediately; *monitored* is what the weekly pass has acted on. Each brand's pass runs on its own cycle, so there is no fixed day to wait for. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No keywords for *your brand* yet. Your portfolio is built during setup and appears here when it is ready." | The keyword list has not been built for your brand. That step is run by hand at EOMA and is not triggered by signing up. | Ask support to start it. Every SEO screen stays empty until then. | | "No keywords match these filters." on a brand that clearly has keywords | The status filter opens on *Monitored*, not on everything. | Switch the status list to **All**. | | "You've pinned as many keywords as your plan tracks. Un-pin one to make room." | Your pins have filled the weekly budget. | Un-pin one, or raise the cap by changing plan on [Plan and usage](/docs/settings/plan-and-usage). | | "Your plan measures N keywords but you have pinned M. The excess will not enter the next scan." | More pins than your plan's weekly budget, usually after a downgrade. We will not name which rows are excluded, because the order is not meaningful. | Un-pin down to the cap so you choose what is dropped, rather than us. | | "Could not add that keyword." / "Could not save that change." | The write failed. Nothing was changed. | Try again. If it repeats, contact support. | | A keyword you added is not in the table | New pins wait in *Measured on the next scan* above the table until the weekly pass picks them up. | Nothing to do. It moves into the table by itself. | | You removed a keyword and want it back | *Remove* hides the keyword and closes the discovery door; it deletes nothing. | Set the status filter to **Set aside** and press **Put back**. | ## Related - [Rank tracking](/docs/pages/rank-tracking) — the weekly measured set, with positions - [Opportunities](/docs/pages/opportunities) — which of these keywords to work first - [Plan and usage](/docs/settings/plan-and-usage) — the pool cap and the weekly cap --- # Rank tracking > The keywords your plan measures on Google every week, where each one sits today, and which ones have never been found at all. Source: https://eoma.ai/docs/pages/rank-tracking Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - This screen is the set your plan measures weekly — the smallest and most expensive of the three keyword counts in EOMA. - Week-over-week movement is not shown yet. The Movement column reads "— new" on every row until the weekly schedule has run more than once for your brand. - A keyword missed on three or more checks is labelled, because it holds a paid slot either way. ## Before you start - Your keyword list must exist and the Google scan (weekly on Growth and Scale, monthly on Starter and the free plan) must have run at least once. Both are set up for you — see [Portfolio](/docs/pages/portfolio). - Nothing else. Every plan opens this screen. Your plan decides how many keywords are in the set: 10 on the free plan, 25 on Starter, 50 on Growth, 100 on Scale. See [Plan and usage](/docs/settings/plan-and-usage). ## Steps ### 1 — Open SEO → Rank Tracking ``` ┌─ sidebar ──────────────┐ │ ● SEO │ │ Portfolio │ │ Rank Tracking ← click │ Competitors │ └────────────────────────┘ ``` ### 2 — Read the distribution bar before the table The bar answers "how am I doing" in one glance. The table answers "on what", which is a slower question. ### 3 — Set aside the keywords that are not your market The last column carries the same **Pin** and **Remove** controls as the Portfolio screen. A keyword that has never been found is still using one of your weekly slots — leave it if you are deliberately aiming for it, remove it if it was never your market. ## What you'll see ### "Rank tracking" and the big "On page 1" figure The title strip, then a count written as *page-1 keywords / monitored keywords*, with a sentence naming both, the top-3 count, and the date of the last check. ### The five-column rail — "Monitored", "Ranking", "Top 3", "Page 1", "Avg position" Under the headline, describing the monitored set only. ### The movement notice A bordered line explaining that week-over-week arrows appear once the weekly schedule has run more than once, and that what you are seeing now is a single snapshot with the date it was captured. ### "Where your monitored keywords sit today" A stacked bar in one panel: Top 3, Page 1 (4–10), Page 2 (11–20), Past 20, Not ranking, each with its count beside it. ### "Monitored keywords" The table: Keyword, Google, Intent, Position, Movement, Yours. Ranked keywords are listed first, best position at the top; the ones that do not rank follow, ordered by search volume. ### The note under the table Present only when keywords have gone unfound for three or more checks. It counts them across the whole set, not only the rows on screen. ## Reading the numbers **"On page 1" — for example 18/50.** Eighteen of the fifty keywords your plan measures are in Google's top ten. The other 32 are being checked every week and are not there yet. **"Monitored — weekly SERP set" — for example 50.** The keywords a full Google results page is fetched for each week. This is the number your plan is priced on. It is not the same as *Measured* on [Share of voice](/docs/pages/share-of-voice) (every keyword we hold any rank for, often several hundred) or *Keywords* on [Portfolio](/docs/pages/portfolio) (the whole pool, often a few thousand). Three counts, three meanings, and only this one is the weekly budget. **"Ranking — of 50" — for example 37.** You appear in the top 20 on 37 of the 50. **"Avg position — where ranked" — for example #8.4.** Averaged over the 37 that rank, not over all 50. Keywords with no position are left out rather than counted as a bad rank, so this number gets *worse* when a weak keyword starts ranking. Read it beside Ranking, never alone. **The distribution buckets.** *Top 3* is won. *Page 1 (4–10)* earns already. *Page 2 (11–20)* is the winnable band — a better page moves these. *Past 20* is shown in the error colour on purpose: a position at #110 is not a weaker page 2, it is a different project. *Not ranking* means the keyword was checked and your domain did not appear at all. **"Last checked 9 Aug 2026" in the headline.** The most recent measurement across the whole set, not today's date. The set refreshes weekly, so a date more than about ten days old means a run has been missed. **The Movement column, reading "— new" on every row.** Movement is not computed on this screen yet. It is not that every keyword is new — it is that the comparison is not built. The notice above the table says the same thing. For the movement that *is* computed, see the *Since the last scan* card on [SEO summary](/docs/pages/seo-summary). **"not found in 6 checks" under a keyword.** That keyword has sat through six weekly full-page checks and your domain has appeared on none of them. It is a statement of fact, not a verdict: the keyword still holds one of your weekly slots every week, and only you know whether it is worth aiming for. The scoring that chose it does not remember these misses, so it will keep being chosen until you set it aside. **A dash in the Position column.** The keyword did not appear in Google's top 20 on the last check, or that check has not covered it yet. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No keywords are in *your brand*'s monitored set yet. We select them from your portfolio, then start tracking their positions." | Either the keyword list has not been built for your brand yet — a step run by hand at EOMA, not triggered by signing up — or it exists and the weekly selection has not run. | Check [Portfolio](/docs/pages/portfolio). If it is empty too, ask support to start the keyword build. | | "Could not load rank tracking." | The read failed. | Refresh; if it repeats, contact support. | | The Movement column never fills in | It is not implemented on this screen yet, whatever the notice above the table suggests about waiting. | Use the *Since the last scan* card on [SEO summary](/docs/pages/seo-summary) for real movement. | | "Last checked" is more than about ten days old | The weekly run has been missed for your brand. | Contact support with the date shown. | | Several keywords say "not found in N checks" | The keyword scoring picks by demand and winnability and does not learn from misses, so an unreachable keyword keeps its slot. | Press **Remove** on the ones that are not your market. That frees a weekly slot for something else. | | A keyword you pinned on Portfolio is not here yet | Pins take effect immediately, but the weekly selection has not run for your brand. | It appears after the next pass. There is no fixed weekday — each brand runs on its own cycle. | ## Related - [Portfolio](/docs/pages/portfolio) — the whole pool, and where keywords are added - [Opportunities](/docs/pages/opportunities) — which of these positions to work first - [Plan and usage](/docs/settings/plan-and-usage) — how many keywords your plan measures weekly --- # Competitors > Choose which rival websites EOMA follows week to week, see where each one outranks you, and read the pages winning the keywords you want. Source: https://eoma.ai/docs/pages/competitors Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - This list is yours to choose. It is not the automatic rivals list on Share of voice, and a 0 here does not mean we found nobody. - Suggested chips are already-measured domains — one click adds one. - Selecting a rival opens two panels: the keywords you both rank on, and their best-performing pages. ## Before you start - Your keyword list and first Google scan must be in place, or there is nothing to compare a rival against. See [Portfolio](/docs/pages/portfolio). - Competitor pages must have been captured on at least some of your keywords for suggestions to appear. That happens on the keywords in your weekly measured set. - Nothing else. Every plan opens this screen, and there is no cap on how many rivals you may follow. ## Steps ### 1 — Open SEO → Competitors ``` ┌─ sidebar ──────────────┐ │ ● SEO │ │ Rank Tracking │ │ Competitors ← click │ Web Analytics │ └────────────────────────┘ ``` ### 2 — Take a suggestion, or type a domain The Suggested row lists domains already measured on your keywords, each with how many keywords it beats you on. Click one to follow it. To add a rival that has not been measured, type its address into the box and press **Track** — `https://` and `www.` are stripped for you. ``` ┌───────────────────────────────────────────┬─────────┐ │ 🔍 Add a competitor domain… │ Track │ └───────────────────────────────────────────┴─────────┘ Suggested + etstur.com +72 + tatilsepeti.com +27 ``` ### 3 — Click a rival to open its detail The first rival is selected for you. Clicking another row, or another bubble in the chart, switches both panels underneath. ## What you'll see ### "Competitive intelligence" and the big "Rivals you're watching" number The title strip, then the count of rivals you follow with *chosen by you — followed week to week* under it, and a sentence naming your biggest threat. With nothing tracked yet, that sentence tells you how many rivals have already been measured and are waiting to be picked. ### The add bar, "Suggested" chips, and "Watching · N" The controls, then the roster header with a dot that reads *live* once the list is in sync across everyone in your workspace. ### "Competitive landscape" A bubble chart. Across: how many of your keywords they share. Up: better average rank. Bubble size: how many of your keywords they beat you on. Your own brand is the mint bubble. Clicking a bubble selects that rival. ### The roster One row per tracked rival: the domain, a High/Medium/Low badge, a bar of how many keywords they beat you on, their keyword overlap, their average rank against yours, and their page count. A ✕ appears on the row when you hover it, to stop following that rival. ### "Head-to-head · *domain*" and "Their top pages" Two panels under the chart, for the selected rival. The first lists the keywords you both rank on, with their position in red where they beat you. The second lists their best pages by estimated traffic; clicking one opens it. ## Reading the numbers **"Rivals you're watching" — for example 0.** Nobody chosen yet. This is not a statement about how much we know — [Share of voice](/docs/pages/share-of-voice) can list six measured rivals while this reads 0, because that list is measured from Google automatically and this one is curated by you. The sentence beside the zero names how many candidates are already measured. **"beats you on 118".** On 118 of your keywords that domain either ranks above you or ranks where you do not appear at all. This is the number that orders the roster and sizes the bubbles — it is the honest "how much is this costing me" figure. **The High / Medium / Low badge.** A fixed banding of that same count: 100 or more is High, 30 or more is Medium, below that is Low. It is a reading aid, not a separate measurement. **"79 kw · avg #6.2 vs #11.4".** They appear on 79 of your keywords, averaging position 6.2 where they do, against your 11.4 across the same kind of set. A rival with fewer keywords but a much better average is a focused competitor; one with many keywords and a worse average is broad but shallow. **"pages 412".** How many pages of theirs we have captured. Useful as a rough sense of site size — a 12-page rival beating you is a different problem from a 400-page one. **A bubble in the top-right corner of the chart.** Wide overlap and good average rank: they compete with you nearly everywhere and win. The bottom-left corner is the opposite. A rival with no average rank at all is pinned to the chart floor rather than being dropped. **Head-to-head — `villa kiralama · Them #3 · You #14 · 6,600`.** They hold position 3, you hold 14, and the keyword sees 6,600 searches a month. Red on their position means they beat you on that row. A dash under *You* means you do not appear there at all, which also counts as a loss. **"Their top pages" — `/antalya-villalari · 3,120 · 47`.** That page is worth about 3,120 visits a month and ranks for 47 keywords. This is the most directly useful thing on the screen: open the page and read what it does that yours does not. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Select a brand to see its competitors." | No brand is active. | Pick a brand from the brand switcher. | | "None picked yet — this list is yours to choose. We have already measured N rivals…" | Candidates exist, none tracked. This is the normal starting state. | Click a Suggested chip. | | "Nobody picked yet — track one above, or take a suggestion." with no suggestions | No competitor page has been captured on your keywords yet, usually because your first weekly full-page scan has not run. | Type a rival's address in manually, or wait for the first scan. | | "That change didn't save — your rival list is unchanged." | The add or remove failed and the list was rolled back. The row you saw appear or vanish did not stick. | Try again, and tell support if it keeps happening. | | "Could not load competitive intelligence." | The read failed. | Refresh; if it repeats, contact support. | | "No shared keywords found." in head-to-head | That domain does not rank on any keyword in your set. Common for a rival you typed in by hand rather than took from Suggested. | Keep it if you know they compete with you; the panels fill as your keyword set grows. | | "No page data yet — runs with the next relevant_pages collection." | We have not captured that rival's pages yet. | Nothing to do. It fills on the next collection. | | The Share of voice rivals list and this roster disagree | They are answering different questions: measured by Google versus chosen by you. | Both are right. Use this screen to decide who is worth following. | ## Related - [Share of voice](/docs/pages/share-of-voice) — the rivals Google shows us automatically - [Opportunities](/docs/pages/opportunities) — the keywords to take back first - [Rank tracking](/docs/pages/rank-tracking) — the weekly set these comparisons rest on --- # Web analytics > Your real website traffic from Google Analytics — who arrived, where they came from, and how many of them were sent by an AI assistant. Source: https://eoma.ai/docs/pages/web-analytics Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-16 --- ## In 10 seconds - This is your own website's traffic, pulled live from Google Analytics — everything else in EOMA measures what AI and search engines *say* about you; this measures who turned up. - It answers where your visitors came from, including how many an AI assistant sent. - Empty until you connect Analytics. There is nothing for EOMA to compute on its own. ## Before you start - **Google Analytics must be connected and a property chosen.** Your *property* is the specific website Analytics is tracking. See [Google Analytics](/docs/integrations/google-analytics). - Nothing else. Search Console is a separate connection feeding a separate screen — see [Search Console](/docs/pages/search-console) — and neither one waits for the other. ## Steps ### 1 — Open it from the sidebar ``` ┌─ sidebar ─────────────────────────┐ │ ▶ SEO │ │ ▼ Web Analytics │ │ ● Google Analytics ← here │ │ Search Console │ │ ▶ Social Listening │ └───────────────────────────────────┘ ``` **Web Analytics** is its own section holding two screens. This page is the first of them; the second is [Search Console](/docs/pages/search-console). It also sits on the dashboard orbit map as the **Analytics** planet. ### 2 — Choose your reporting window ``` ┌───────────────────────────────────────────────┐ │ Web Analytics [24h][48h]│[7d][30d][90d] │ │ example.com · 7d overview ↑ pick one │ └───────────────────────────────────────────────┘ ``` The line divides two kinds of window. **24h and 48h** report hour by hour; the rest report day by day. That changes the grain of every number below, which is why the divider is there. ### 3 — If you have more than one property, pick it When your Google account has several properties linked to this brand, the site name under the title becomes a dropdown. One property, and it is plain text. ## What you'll see Blocks appear only when they have data, so a small site shows a shorter page. Nothing is hidden from you deliberately. ### The header — "Web Analytics" Sticky at the top. Under it: the property name, the window you picked, and — on the 24h and 48h windows only — which timezone the hours are counted in. That last part matters: "nothing since 14:00" is a different fact in Istanbul than in Berlin. ### The four tiles — Active Users, Sessions, Bounce Rate, Avg Duration The headline row, each with a small trend line and a comparison to the window before. Revenue tiles join them if your site reports sales to Analytics, then a row carrying a tag for how Analytics reads your site — E-commerce, Lead Gen or Content — and any sharp movement worth noticing. ### "Sessions & Users" and the live panel The main chart on the left. On the right, **Live**: who is on your site in the last 30 minutes, with the pages, sources and countries they came from. ### "Where Visitors Came From" The widest table on the page, and usually the reason people open it. One row per source, with a bar strip showing when they arrived, a session count, and how long ago the last one was. A row can carry an **AI** tag, and a line above the table totals those. ### "Traffic Sources" and "Top Pages" Side by side. The channel mix on the left is clickable — clicking a channel filters the table above it to the sources behind it. Under them, "Devices" and "Countries", then "New vs Returning Users" and "Landing Pages". ### Conditional blocks Campaigns, Site Search Terms, Google Ads, Ecommerce and Conversion Funnel load a moment after the rest, and only when your Analytics property reports them. ## Reading the numbers **Active Users vs Sessions — for example 1,240 users and 1,830 sessions.** 1,240 different people visited, and between them they came 1,830 times. The gap is repeat visits. Users going up while sessions stay flat means you reached more people who each looked once; the reverse means the same audience came back more often. **Bounce Rate — for example 41.2%, and Avg Duration — 1m 48s.** Four in ten visits ended without the person doing anything Analytics counts as engagement, and the average visit lasted under two minutes counting everyone who left instantly. On the bounce tile, down is good and the colour is inverted to say so. Read the two together rather than alone: both moving is a real change in visitor quality, one moving alone is usually just a change in traffic mix. **"142 sessions came from AI assistants".** 142 visits arrived from a link inside an AI answer — someone asked ChatGPT or Perplexity something, got your site as a source, and clicked. This is the closest thing you have to a direct receipt for the AI-visibility work in the rest of EOMA. Treat it as a floor rather than a full count: not every assistant identifies itself when it sends a visitor, so the true figure is at least this. **"Direct / no referrer".** No source was recorded. Some of it is people typing your address, some is a browser stripping the referrer, and some is an app that does not pass one. It is the one row you cannot draw conclusions from. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Connect Google Analytics" with a three-step list | Analytics has never been linked for this brand. | Follow the three steps, or read [Google Analytics](/docs/integrations/google-analytics) first. | | "Finish Connecting Google Analytics" | The Google sign-in started but the analytics permission was never approved, so no lasting access was granted. | Run the connection again and approve every permission it asks for. | | "Pick a Property" | Your Google account is linked but no property is chosen for this brand. | Open Settings → Integrations and pick the property that tracks this website. | | "Reconnect Required" | Google revoked or expired the authorization — most often a password change or a permission removed on Google's side. | Reconnect from Settings → Integrations. Nothing is lost; collection resumes. | | "No Access to This Brand" | Your EOMA account cannot view this brand's analytics. Reconnecting will not fix it. | Ask whoever owns the brand to add you. | | "Couldn't Load Analytics" with a Retry button | A transient failure — network or a Google hiccup. | Press Retry. It usually works the second time. | | You came looking for search keywords and there are none here | Search Console moved to its own screen; this page is Analytics only. | Open [Search Console](/docs/pages/search-console) in the same sidebar section. | | The whole page is thinner than yesterday | Blocks render only when they have data in the window you picked. A short window on a quiet site drops several. | Switch to 30d or 90d before concluding anything is missing. | | Hour-by-hour numbers look shifted | Analytics counts hours in the *property's* timezone, which may not be yours. The header names it on those windows. | Read the timezone note in the header before comparing to your own clock. | ## Related - [Google Analytics](/docs/integrations/google-analytics) — connecting it, and choosing a property - [Search Console](/docs/pages/search-console) — the other screen in this section: what people searched before they saw you - [SEO summary](/docs/pages/seo-summary) — where you stand in search, without needing either connection --- # Search Console > What people typed into Google before they saw you — every query, page, country and device, across five Google search surfaces. Source: https://eoma.ai/docs/pages/search-console Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-16 --- ## In 10 seconds - This is Google's own record of you appearing in search results — what was typed, which of your pages came up, and how high. - It is a separate connection from Google Analytics. Having one never gives you the other. - Every number here lags two to three days behind today. That is Google's delay, not EOMA's. ## Before you start - **Google Search Console must be connected and a site chosen.** See [Search Console](/docs/integrations/search-console) for the connection itself. - Your site must already be verified in Google's own Search Console before EOMA can read it. EOMA reads what Google reports; it cannot grant itself access. - Nothing else. Google Analytics is not required — the two screens are independent. ## Steps ### 1 — Open it from the sidebar ``` ┌─ sidebar ─────────────────────────┐ │ ▶ SEO │ │ ▼ Web Analytics │ │ Google Analytics │ │ ● Search Console ← here │ │ ▶ Social Listening │ └───────────────────────────────────┘ ``` **Web Analytics** is its own section, and Search Console is the second of its two screens. The first, Google Analytics, answers a different question — see [Web analytics](/docs/pages/web-analytics). ### 2 — Pick your window ``` ┌──────────────────────────────────────────────────────┐ │ Search Console example.com Owner [7d][30d][90d][16m] │ └──────────────────────────────────────────────────────┘ ``` Four windows, all counted in whole days. There is no hourly view here because Google does not report hourly. **16m** is sixteen months, the furthest back Google keeps this data. ### 3 — Choose the search surface The row directly under the title — Web, Images, Video, News, Discover. It changes what every number on the page is *about*, so treat it as a mode rather than a filter. ### 4 — Open a card, or a tab The Overview tab shows every dimension as a collapsed card: one big number and its top few rows. Click a card to expand it into the full table, or use the tab strip to go straight to one. ## What you'll see ### The header — "Search Console" The site address, a badge for your access level on it (**Owner**, **Full access**, **Restricted** or **Unverified**), and the window pills. A **Partial** badge appears beside them when the most recent day or two is still being counted by Google and may still change. ### The surface row — "Web · Images · Video · News · Discover" Its own row, above the numbers it governs. ### The four totals — Clicks, Impressions, CTR, Avg. position On one raised panel, each with its change against the window before. ### The tab strip — "Overview · Queries · Pages · Countries · Devices · Appearance" Some tabs grey out on some surfaces. That is deliberate: Discover has no concept of a search query, and an empty Queries card would read as "you rank for nothing" rather than "this is not a search". Hover a greyed tab and it tells you which surface and dimension do not go together. ### The cards, on Overview One per dimension: a tinted head, one big number, then the top rows with a bar showing how much of the total each took. The bar is the point — it shows whether one query carries you or a hundred share the load. ### The full table, inside an opened card Query, Clicks, Impressions, CTR and Position, with **Load more** at the foot. Queries and Pages also get a filter box. A card fetches its table the first time you open it and never again, so the first open is the slow one. ## Reading the numbers **Clicks and Impressions — for example 380 clicks from 24,900 impressions.** Your pages were shown in Google's results 24,900 times, and 380 of those turned into a visit. Impressions are demand you are *already* visible for; clicks are what you converted from it. The gap between them is the cheapest thing to improve in the whole product — you do not need a better ranking to win it. **CTR — for example 1.5%.** Of everyone who saw you in results, fifteen in a thousand clicked. A low CTR on a query with plenty of impressions is almost always a title-and-description problem, not a ranking one: Google is already showing you and people are choosing someone else's wording. **Avg. position — for example 12.4.** On average you sat twelfth. This is the one metric where **down is better**, and the change pill beside it is inverted to say so. Past roughly ten means page two, where clicks collapse — so 12.4 → 9.4 is worth far more than 32 → 22, even though both move ten places. **A query with impressions but no clicks.** Google thought you were relevant enough to show and nobody picked you. Read the page it points at: usually you rank for a question your page does not visibly answer in its title. **The share bar in a card's top rows.** How concentrated that dimension is. Three queries carrying most of your clicks is fragile — one ranking change takes the lot. A long flat spread is slower to build and much harder to lose. **The Partial badge.** The most recent day or two is still being counted and *will* change. Never read a fall in the last day as a real fall; compare complete days only. **"Restricted" on the access badge.** Google is giving EOMA a limited view of this site, so some rows may be missing. Ask whoever owns the Search Console property to raise EOMA's access to Full or Owner. ## When it goes wrong | You see | Why | What to do | |---|---|---| | A message to connect Search Console | It has never been linked for this brand. | Connect it — see [Search Console](/docs/integrations/search-console). | | "Pick a Search Console site" | Connected, but no site is chosen for this brand. | Open Settings → Integrations and pick the site. | | A reconnect prompt | Google expired or revoked the authorization, most often after a password change. | Reconnect from Settings → Integrations. Nothing collected so far is lost. | | An access-denied message | Your EOMA account cannot view this brand's data, and reconnecting will not change that. | Ask whoever owns the brand to add you. | | Tabs greyed out | The surface you picked cannot answer that dimension — Discover has no queries or devices. | Switch back to **Web**, or read the tabs the surface does support. | | "No data in this window" | Real, and it means Google recorded nothing here. | Widen to 90d or 16m before concluding anything. A quiet week on a small site is normal. | | Numbers lower than Google's own Search Console | You are almost certainly comparing different windows or a different surface. | Match the window and set the surface to Web, which is what Google shows by default. | | Today missing entirely | Google reports on a two-to-three day delay, always. | Nothing to fix. Yesterday is expected to be absent too. | ## Related - [Web analytics](/docs/pages/web-analytics) — the other half of the section: who actually arrived, from every source - [Search Console](/docs/integrations/search-console) — connecting it, and choosing a site - [SEO summary](/docs/pages/seo-summary) — where you stand in search without needing any connection --- # Social listening > What people say about your brand on social, read the way an AI engine reads it — what is at risk, what is working, and who carries the most weight. Source: https://eoma.ai/docs/pages/social-listening Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 Available on: growth, enterprise --- ## In 10 seconds - This is not a posting tool. It reads what other people wrote about you and tells you what it means for how AI answers about your brand. - Twitter / X is the platform with real data today. LinkedIn has its own view. Reddit, Instagram and TikTok are marked **Soon** and show a placeholder. - It is a **Growth** feature. On Starter you see a lock rather than an empty page. ## Before you start - Your brand must be on **Growth or higher**. On Starter and on the free plan this page shows a lock. See [Plan and usage](/docs/settings/plan-and-usage). - Your first collection must have finished. Until then you get "We're still preparing this view" — that is timing, not a restriction. - Nothing to connect. EOMA collects public posts itself; you do not link a social account for this — and it does not publish to social accounts at all, so there is no account to link for that either. ## Steps ### 1 — Pick a platform in the sidebar ``` ┌─ sidebar ────────────────────────────┐ │ ▼ Social Listening │ │ ● Twitter / X ← here │ │ LinkedIn │ │ Reddit [Soon] │ │ Instagram [Soon] │ │ TikTok [Soon] │ └──────────────────────────────────────┘ ``` The three with a **Soon** badge open a placeholder card, not data. Clicking the section header alone lands you on Twitter / X. ### 2 — Stay on Topics unless you want the raw stream ``` ┌──────────────────────┐ │ [ Topics ] [ Feed ] │ ← Topics is the analysis └──────────────────────┘ ``` **Topics** is the read: sentiment, risk, who is loudest, what changed. **Feed** is the unprocessed timeline. The toggle does not appear on LinkedIn or on a **Soon** platform. ### 3 — Click anything to see the posts behind it A donut segment, a topic, a voice, a point on the trend — each one opens a panel listing the actual posts that produced the number. Nothing here is a figure you have to take on trust. ## What you'll see ### The heading — "What people *say* about *your brand*" A short intro band across the top, then an explainer card you can dismiss. ### The section title — "How AI sees you on Twitter" With a line explaining the premise: what people say about you in public is the raw material AI engines draw on when they answer questions about your brand. ### The sentiment donut and the three figures beside it A ring split into Positive, Neutral and Negative with the negative share called out in the middle, and to its right: **AI-risk score**, **Negative reach** and **Classified mentions**. ### "At-risk conversations" and "Positive signals" Side by side. The left lists negative posts about you, most serious first, each with a severity tag and a confidence figure. The right lists favourable posts ranked by how far they reached. ### "Topics by sentiment" and "Top voices" Side by side under those. Topics ranks the subjects people raise by how negative the conversation is. Top voices ranks the accounts by audience size, with how each one feels about you. ### "Sentiment over time" A full-width chart at the bottom, weekly volume by sentiment. ## Reading the numbers **AI-risk score — for example 34 / 100.** Of all the audience that has heard something about you on this platform, 34% of that reach was negative. It is weighted by follower count on purpose: one complaint from an account with 200,000 followers counts for more than fifty from accounts with 40, because that is the weighting an AI model effectively applies too. The colour moves with the number — green under 30, amber to 60, red above. **Negative reach — for example 84.2K, "vs 12.1K positive · louder than positive".** The critical posts about you reached about 84,200 people; the favourable ones reached 12,100. When negative is louder, the version of your story with the widest circulation is the unflattering one, and that is the version an AI model is most likely to have absorbed. The fix is rarely to argue with the critics — it is to give the favourable side more reach. **Classified mentions — for example 612.** EOMA found and read 612 posts naming your brand in this window. This is the sample size behind everything else on the page: at 612 the topic breakdown is worth acting on; at 12 it is an anecdote. **"38% neg · 44" on a topic row.** Forty-four posts raised this subject and 38% of them were negative. Topics are sorted worst-first, so the row at the top is the angle most likely to surface unfavourably when someone asks AI about you. Its position matters more than its exact percentage. **"120K followers" on a top voice.** That account reaches roughly 120,000 people when it mentions you. A single negative voice at that size outweighs a long tail of small positive ones in what AI ends up learning — which is why this list is ranked by reach and not by how often someone posts. **"confidence 87%" on an at-risk post.** How sure the classifier is that this post is negative about *you* specifically. Anything low is worth opening before acting on — sarcasm and brand names that double as ordinary words are where it slips. **A spike in the weekly chart.** One bad week is noise. Several negative weeks in a row is what actually reshapes how AI describes you, because models learn from the accumulated conversation rather than from any single post. ## When it goes wrong | You see | Why | What to do | |---|---|---| | A lock and "Upgrade to Growth" | Your brand is on Starter, which excludes social listening. Nothing is broken. | Upgrade from Settings → Plan and usage. Collection starts the moment you do. | | "We're still preparing this view" | Your first collection has not finished. This is timing, not your plan. | Come back later. Nothing to fix. | | "No brand mentions found" on Twitter / X, on a Growth plan | EOMA scanned and nothing surfaced. For a small or very local brand this can be genuinely true. | If your brand is regularly mentioned in public and this persists past a week, report it to support — on a Growth plan an empty Twitter view is worth checking rather than accepting. | | "Coming soon — Instagram intelligence is on the way" | That platform's collection is not built yet. The **Soon** badge in the sidebar says the same thing before you click. | Use Twitter / X. Nothing you can configure changes this. | | An unfamiliar address like `/social/youtube` lands you on Twitter | Only five platforms exist here. An unknown one redirects to the default rather than showing a blank page. | Pick from the sidebar. | | LinkedIn has no Topics / Feed toggle | LinkedIn has its own single surface rather than the two-view split. | Expected. Everything LinkedIn offers is on the one screen. | | A post that is not about your brand at all | Brand names that are also ordinary words collect false matches. | Note it and tell support which post — the classifier is tuned centrally, not per brand. | ## Related - [AI provider visibility](/docs/pages/ai-provider-visibility) — what AI actually says, once it has read all this - [Plan and usage](/docs/settings/plan-and-usage) — which plan includes social listening - [Actions](/docs/pages/actions) — where a negative topic becomes work EOMA can do --- # Actions > This week's list of fixes as one table, cut by whose turn it is, with one button that hands a job to EOMA and a way to do it yourself. Source: https://eoma.ai/docs/pages/actions Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-09 --- ## In 10 seconds - The list is one table you filter by **whose turn it is**: what waits for you to start, what EOMA is working on, and what waits for your review. - **Start** hands one job to EOMA, and there is no way to un-start it. It uses one of the jobs your plan has EOMA do for you; you can also do any action yourself, with its steps. - Nothing EOMA writes reaches your website until you approve it, and on a code repository you still merge it yourself. ## Before you start - **Your first collection must have finished.** Until then the page says your queue is still being prepared. There is nothing to fix and nothing to press. - **Nothing else is required to read the list.** What is on it depends on your plan: on Starter, six fixed tasks, each coming back on its own schedule; on Growth and above, also new actions every week from what EOMA measured. - **To have EOMA deliver finished work to your site, connect somewhere for it to land** — WordPress or GitHub. Without a connection EOMA still prepares the work; you take the file and place it yourself. See [Integrations](/docs/settings/integrations). ## Steps ### 1 — Open the list Sidebar → **Actions**. That lands you here; there is no menu in between. ### 2 — Read the tabs that have your name on them The bar above the table cuts the same list by whose turn it is. Two of the tabs are yours: **Waiting for you to start** holds the actions EOMA has prepared and nobody has started, and **Waiting for your review** holds everything that cannot move until you decide something. The number beside each tab is how many rows are in it. On a phone there are no tabs: the deck shows everything still open (see "The columns" below). ### 3 — Start one job you want EOMA to do Each row nobody has started has a **Start** button. Pressing it opens a confirmation that says what happens and what it uses on your plan: ``` ┌─ Start this action? ────────────────────────────┐ │ EOMA starts working on it and comes back with │ │ a draft in about 2 days; nothing is published │ │ until you approve it. It uses one of the 2 jobs │ │ EOMA does for you on your Free plan; 1 will be │ │ left. │ │ │ │ [ Cancel ] [ Yes, start ] │ └─────────────────────────────────────────────────┘ ``` There is no un-start, no refund, and no cancel once EOMA has the job. If you would rather do that one yourself, open it and use its steps (see "The reading panel" below), or use **"I've already done this"** under **Remove**. ## What you'll see ### One table, its tabs Every action is one row. The tabs above it are filters, not different screens, in the order an action walks them: | tab | what is in it | |---|---| | **All** | everything still open | | **Waiting for you to start** | actions nobody has started — press **Start**, or do one yourself with its steps | | **EOMA does these** | actions EOMA is working on right now | | **Waiting for your review** | needs a decision from you — a draft to approve, a question to answer, a pull request to merge | | **Archive** ▾ | opens **Completed** and **Removed**, both closed by default | Removing is never a delete. Anything you take off the board waits under **Removed**, struck through, and goes back with one press. ### The columns **Action · Type · Status · Priority · Added · Next move.** **Type** is which part of your marketing the row is about — *AEO* (how AI engines answer about you), *SEO*, *Blog*, or *Web* (your site itself). **Next move** is the only column you press. It holds one control per row, the one that fits that row right now: Start, Remove, Approve, Answer, or nothing at all while EOMA is working. On a phone the page changes shape. The title **Actions** carries the *ETA ~2 days* badge and a **?** button that opens **How actions work**; the **Documentation** link is at the foot of those screens. Under the title stand your figures and your agents. If no site is connected yet, the row **Let EOMA apply fixes to your site** comes next (under the deck on a small screen). Then the deck: every open action, one per card. Swipe a card left for the next action and right for the previous one, or tap the right or left edge of its picture; after the last card the first comes back. Each card has the same buttons as its row, and tapping the card opens the action. A card shows its status at the top left of its picture, and it does not say how long ago the action was added. A phone has no tabs and no **Archive**: the deck always holds everything still open, and pressing a free agent turns it to the next action waiting for you to start. **Completed** and **Removed** are on a computer or a tablet. **Status** names where the row stands. A row nobody has started says **Waiting for you to start** — on every plan, also once the jobs your plan has EOMA do are used, because starting it is still up to you. **Priority** is drawn as three rising bars beside the level's word: one bar lit is **Low**, two is **Medium**, three is **High**. **Critical** is a red pill with a warning sign — start with those rows. On a phone the same mark sits at the right of the card, above its title. ### The two marks that report state **Your agents, top right, and the agent at the start of a row.** Each done-for-you job your plan gives you is drawn as one of EOMA's agents, busy ones first. An orange agent is working on a row; a teal one is waiting for you to answer or approve; a framed one is free; a faded one with its eyes shut is a job your plan has used for good (Free and Starter count jobs for the life of the plan). The line beside them says how many are free, or how many wait for you. The same agent stands at the start of the row it is on, so you can see where each one is. Press a busy agent to open its row, or a free one to see the actions waiting for you to start. **Three dots under the status.** They drift left and right while the row waits on someone: on **you** to answer, approve or merge, or on EOMA while it works. The chip above them says which. **The ETA badge.** In the list's header, *ETA ~2 days* says how long an action takes on average. Hover over it or press it to see why that can vary: each time EOMA needs information from you, or a sensitive action waits for your approval, the work takes one more round. It is a typical figure, not a countdown. ### The reading panel Clicking any row opens it over the board. It holds what EOMA prepared, then a **Your move** block with what fits that row: how a draft should end (publish it, take it away, or request changes), your answer when EOMA asked something, the pull request to review, or Mark done. When EOMA needs something from you, the block says exactly what under **We need something from you**, in EOMA's own words, and holds the answer form: type the answer, attach a file, or both. Once EOMA has prepared something, the last step, where you choose how it ends, also lets you take it away as PDF, Markdown or HTML, or copy it as a prompt for your own AI. On an action nobody has started there is nothing to export yet. Instead, a **Doing it yourself?** box sits beside the explanation. **Copy the prompt** puts full instructions for this action on your clipboard, ready to paste into ChatGPT or Claude. When you have done the work, press **I applied this myself** and confirm. Pressing **Start** at the bottom hands the job to EOMA instead. Beside it, one line says what the press does and what it uses on your plan — for example, *"Press Start and EOMA does it; the draft comes to you for approval in about 2 days. On your Free plan EOMA does 2 jobs for you."* On a phone that line sits above the button. ### "How actions work" The first button in the board's toolbar, with a small book symbol. It opens seven short screens beside the table that explain where the suggestions come from, why a full board stops receiving new ones, what Start spends, and why your approval is the last gate before anything reaches your site. It is the same explanation this page gives, drawn rather than written. On a phone it is the **?** beside the title, and the screens open above the deck. ## Reading the numbers **The count on each tab.** How many rows that filter holds right now. Two are worth watching. **Waiting for your review** — while it is above zero, EOMA is stopped on something only you can unblock. **Waiting for you to start** — EOMA does nothing with these until you press Start, or until you do them yourself. **"Your list is full."** Your board holds a set number of open items: **six** on Starter, **ten** on Growth, **twelve** on Scale. It is a limit on the pile, and it is not a queue. While the board is full, **new suggestions do not land**, so a full list is a stopped conveyor belt rather than a backlog waiting for you. Finishing one or removing one starts it again. **"On your Growth plan EOMA runs 3 of your jobs at a time; all three are running now. When one finishes you can start this one."** Start is refused because every place your plan gives you is held by a job already in flight. Nothing resets on a schedule: a place frees the moment one of those jobs reaches done, or you remove it. Everything else keeps working — you can still read, remove and mark actions done. **"On your Starter plan, all of the 3 jobs EOMA does for you have been used."** On Free (2) and Starter (3) the jobs EOMA does for you are counted for the life of the plan and do not come back when a job finishes. The refusal says you can open the action and do it yourself with its steps, and carries **See plans**, which opens the plan choice; upgrading takes one more press there. **"Taken on 4 days ago · no update yet".** EOMA holds that job and has reported no progress for four days. It still holds its execution. One or two days is normal for work that needs research; a week is worth reporting. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "…EOMA runs 3 of your jobs at a time; all three are running now." | Every place your plan gives you is held by a job in flight. On Growth that is three at a time; nothing frees on a schedule. | Finish one of the jobs in flight, or remove one you no longer want — a place opens the moment you do. Or upgrade for more at once. Everything else on the page keeps working. | | "…all of the 3 jobs EOMA does for you have been used." | On Starter the jobs EOMA does for you are counted for the life of the plan. It shows at **Start**, or inside the confirmation, under a warning sign, when the jobs were used in another tab or on another device while this page was open. | Do the action yourself with its steps, or press **See plans** to upgrade. | | Your waiting cards show a lock and **Upgrade** | On Free, EOMA does 2 jobs for you for the life of the plan. Once both are used — finished ones included — the other cards wait behind the lock. | Press **Upgrade** to have EOMA do more. | | Nothing new arrived this week | The board is at its ceiling (six open items on Starter, ten on Growth, twelve on Scale), and nothing new lands while it is full. On Starter, a fixed task also comes back only on its own schedule. | Finish one or remove one. The next batch arrives on its own after that. | | "Your action queue is still being prepared" | The week's list has not been generated for this brand yet. | Come back later. This usually resolves within an hour of brand setup finishing. | | "Couldn't load your actions. Please refresh." | The read failed. Your actions are not lost. | Refresh. If it repeats, contact support. | | "This action was just started; EOMA is working on it." | Something started it first — another browser tab or device. | Press **Close**. The action is in progress; nothing was used twice. | | "The action couldn't be started right now. None of your jobs was used; try again in a little while." | The start did not get through — a dropped connection or a server error. | Press **Try again**, or come back later. | | "This action can no longer be started; your board has been updated." | The action left the board in the meantime, or it is not a job Start hands to EOMA. | Press **Close**. Your board now shows its latest state; none of your jobs was used. | | "Your plan has no jobs left for EOMA to do for you." | Every job your plan gives EOMA has been used; the screen had not shown the count yet. | Press **Close**. The **Plan & Usage** page shows how many you have left. | | "Another tab beat this click. Refresh to see the latest." | Two tabs acted on the same row. | Refresh. Nothing was applied twice. | | A row EOMA holds says it is waiting on a connected site | EOMA finished preparing and has nowhere to deliver it. | Either connect a site in Settings → Integrations, or open the row, export the draft, and place it yourself, then mark it done. | | A row you started sits with no update for over a week | The job is claimed but nothing is moving on it. | Contact support with the action's title. The execution it holds will not free on its own. | ## Related - [Life of an action](/docs/concepts/life-of-an-action) — every state an action passes through, and who is blocked at each one - [Plan and usage](/docs/settings/plan-and-usage) — how many done-for-you seats your plan gives you - [What EOMA does and what you do](/docs/start-here/what-eoma-does-and-what-you-do) — the split this page is built on --- # Brand index > The canonical record of facts about your brand that AI engines read — confirm what EOMA found, correct what it guessed. Source: https://eoma.ai/docs/pages/brand-index Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - One record per brand: your name, what you do, where you are, how to reach you — written so a machine can read it. - **To confirm** holds the facts EOMA guessed rather than found. Your answer always replaces ours. - Your correction is live on the public record from the next request. There is no publish step to press afterwards. ## Before you start - **Your brand must be matched to a record.** EOMA matches by the web address on your brand — no address, no match. See [Brand profile](/docs/settings/brand-profile). - **The record must have been collected.** EOMA builds it by reading your website. Until that has run, the page says no record was found for this brand. - Nothing else. There is no plan restriction on this screen and correcting your record costs nothing. ## Steps ### 1 — Open Brand Index Sidebar → **Brand Index**. It is its own entry; there is no menu in between. If the page says no record was found for this brand, the steps below will not work yet — see "Before you start". ### 2 — Work through "To confirm" The button at the top right, **Review 4 flags**, scrolls you straight to the queue. Each entry is one field EOMA was not confident about, with a short reason and two choices: ``` ┌─ Brand name ────────────────────────────┐ │ We guessed this from your site text. │ │ [ Confirm ] [ Edit ] │ └─────────────────────────────────────────┘ ``` **Confirm** accepts EOMA's value as correct. **Edit** replaces it with yours. Either way the entry leaves the queue and the field becomes part of the record. ### 3 — Mark the record ready for review When the queue is empty, the top-right button becomes **Mark ready for review**. Press it when you are satisfied with the whole record. It tells the EOMA team the record is finished from your side; it is not what makes your edits take effect — those are already in force. ## What you'll see ### The name, and "This is the canonical record AI engines cite when they answer questions about your brand." Your brand name in large type, your domain, and **Updated** with the date the record last changed — including by your own edits. ### The status pill — Seed · Draft · Ready for review · Published Top right, beside the button. **Seed** is a record EOMA collected and nobody has touched. **Draft** means you have edited or confirmed at least one field. **Ready for review** means you pressed the button. **Published** is set by the EOMA team, not by this screen. ### The three-panel scoreboard — "Confirmed", "To confirm", "Completeness" A single strip under the header. Read it left to right: what is settled, what still wants you, and how much of the record exists at all. ### "To confirm" The queue, with a coloured spine per reason: **Inferred**, **Low confidence**, **Title junk**, **Objectivity**, **Missing**. Each row states the question in EOMA's own words, then why it is being asked, then Confirm and Edit. When it is empty it says so plainly: *"Nothing to confirm — every field is sourced or confirmed."* ### "Record" Everything already settled, one row per field: the field name on the left with a small source code under it, the value on the right. Fields you edited are marked **Edited by you**. ### "Guide" and "Reach" (right column) **Guide** repeats your progress as a bar and states the three rules of this screen: AI engines cite this record, your edits always win over EOMA's, every fact shows its source. **Reach** collects the contact details and profile links in the record — your website, support email, phone, languages, and the other places your brand appears. ## Reading the numbers **Confirmed — for example 12.** Twelve fields in the record are either sourced from something EOMA trusts or confirmed by you. These are the facts an AI engine would repeat about your business today. **To confirm — for example 4.** Four fields are waiting on a human answer. Each one is a fact EOMA is not confident enough to stand behind, so it is showing you its guess rather than publishing it quietly. Four is a ten-minute job, and it is the highest-value ten minutes on this screen. **Completeness — for example 75%.** The share of the record that is settled rather than outstanding — the confirmed count as a proportion of confirmed plus outstanding. It is **not** a quality score and not a comparison against other brands: it reaches 100% the moment your queue is empty, whether the record is rich or thin. A thin record at 100% and a detailed one at 100% look identical here, so use the Record panel, not this number, to judge whether the facts are good enough. **"9 of 13" in the Guide, and "Confirm 4 more to be ready for review."** The same arithmetic said as a next step. When it flips to *"All fields confirmed"*, the Mark-ready button is the only thing left to press. **The small code under a field name — `og`, `dom`, `jsonld`, `wikidata`, `llm`.** Where the value came from, in EOMA's own shorthand: tags in your page's source, the visible text of the page, Wikidata, or a model reading your site. It is engineering shorthand that has not been translated yet, and it is listed here in descending order of how much EOMA trusts it. A field you changed shows **Edited by you** instead, which outranks all of them. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Your brand isn't indexed yet" | Your brand has not been matched to a record. The match is made from the web address saved on your brand. | Check the address in Settings → Brand profile. The link is then made automatically — there is no button. | | "No record found for this brand yet" | Your brand is matched, but the collection that builds the record has not produced one. | Wait. It appears here as soon as the collection runs; nothing on your side is missing. | | A field you corrected still looks wrong on your public record | Your edit is applied on the next request, but a page you already have open, or a cache in front of it, may still hold the old copy. | Reload the public page. If it is still wrong after a reload, contact support with the field name. | | A question in the queue is written in English on a Turkish workspace | The questions are written when the record is collected and are stored in English. This is a known gap, not a fault in your record. | Answer in whatever language your customers use. Your answer is stored exactly as you type it. | | "Mark ready for review" is greyed out | You already pressed it — the record is at Ready for review or Published. | Nothing to do. It stays that way; you do not need to press it again. | | The Record panel says nothing beyond the essentials yet | EOMA found only a few facts it trusts on your site. | Answer everything in To confirm — confirmed fields move into the Record. Adding clear facts to your own website is what grows it further. | ## Related - [Actions](/docs/pages/actions) — this week's work on the facts you confirm here - [Brand profile](/docs/settings/brand-profile) — the web address that connects your brand to its record - [Your brand on Google](/docs/pages/your-brand-on-google) — what AI currently says about you --- # What's new > Every feature added to EOMA, newest first. Where to check what changed before you assume a screen is broken. Source: https://eoma.ai/docs/pages/whats-new Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - A single list of everything shipped, newest at the top. Each headline expands to a short paragraph. - Opening the page marks everything read. There is no "mark as read" button to press. - The card on your dashboard is a pointer at this page, not a second copy of it. ## Before you start Nothing — this page works from your first day. It is the same list for every brand and every plan, so nothing here depends on your data or your subscription. ## Steps ### 1 — Open it from the bottom of the sidebar ``` ┌─ sidebar ─────────────┐ │ … │ │ ───────────────── │ │ Wish box │ │ ● What's new ← here │ │ Settings │ └───────────────────────┘ ``` It sits directly under the wish box on purpose — what you asked for, then what was shipped. ### 2 — Click a headline to read the detail ``` ┌───────────────────────────────────────────────┐ │ A shorter path from an action to a draft ▾ │ ← click anywhere │ Aug 11 │ └───────────────────────────────────────────────┘ ``` The row expands in place. Click it again, or open another, to collapse it. ### 3 — Notice the dot on the dashboard card The **What's new** card in the bottom-left of the dashboard carries a small coloured dot when something has shipped since you last opened this page. Opening the page clears it. ## What you'll see ### The heading — "What changed in EOMA" With the line "Every feature we've added, newest first" under it. ### The list One row per change: the headline on the left, the date on the right, and a chevron. Rows are separated by hairlines rather than sitting in separate cards. There is no paging — the whole history is on the page. ### The expanded paragraph A few sentences under the headline you clicked, in plain language. No version numbers, no technical detail. ## Reading the numbers There is only one figure on this page. **The date beside each headline — for example "Aug 11".** The day that change went live for everyone, not the day it was written or merged. It is the number to check when a screen behaves differently from how you remember it: if a change landed the day before you noticed, the two are almost certainly the same event, and there is nothing to report. **The dot on the dashboard card.** Something shipped after the last time you opened this page. It is per person, not per brand — clearing it on your account does not clear it for a colleague. Closing that card with its **×** hides it until something newer than the entry you closed on ships, so you will not lose an announcement by dismissing the card. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Nothing here yet" | No entries have been published. On a new workspace this is normal for a short while. | Nothing to fix. Check back after a release. | | "We couldn't load what's new" | The list failed to fetch. It says nothing about your brand or your data. | Refresh in a moment. If it repeats, contact support. | | The dashboard card is gone but entries are unread | You closed the card with its **×** at some point. That hides it until something newer ships. | Open this page from the sidebar. The card is a shortcut, never the only way in. | | Entries in a different language than the rest of the app | Each entry is written in both English and Turkish and follows your interface language. | Change the language from the menu under your name in the sidebar. | | A feature you use is not mentioned here | Not every change is announced. Small fixes and internal work do not get an entry. | If a screen behaves in a way you cannot explain, contact support rather than assuming it was an unannounced change. | ## Related - [Dashboard](/docs/pages/dashboard) — where the What's new card sits - [What is EOMA](/docs/start-here/what-is-eoma) — what the product does, if a headline assumes context you do not have --- # Google Analytics 4 > Connect the Google account that owns your GA4 property, then pick the property — both steps, or the connection returns nothing. Source: https://eoma.ai/docs/integrations/google-analytics Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Connecting is **two steps**: sign in with Google, then choose which GA4 property belongs to this brand. Stopping after the first step leaves you with a connection that reads nothing. - Sign in as the Google account that can already open your GA4 property. If you can't see the property in Google's own Analytics, EOMA won't see it either. - EOMA only reads. The permission it asks for cannot change or delete anything in your Analytics. ## Before you start - You must be the **owner** of this brand in EOMA. Anyone else sees an "Owner only" label instead of the Connect button. See [Integrations](/docs/settings/integrations). - You need to know **which Google account can open your GA4 property**. This is the one that trips most people up: the property was often created by an agency, a developer, or an old company address, not the account you use for email. Open `analytics.google.com` in another tab first and check which account is signed in there — that is the account EOMA needs. - Your *property* — the thing in Google Analytics that collects one website's traffic — must already exist. EOMA connects to a property; it cannot create one. - No plan requirement. Integrations are available on every plan. ## Steps ### 1 — Open Settings → Integrations ``` ┌─ Settings ───────────────────┬─────────────────────────────────┐ │ Workspace │ Analytics & Search │ │ Plan & Usage │ │ │ Brand Profile │ ▮ Google Analytics 4 [Connect]│ │ ● Integrations ← here │ ▮ Google Search Console │ │ Preferences │ ▮ Google Drive │ └──────────────────────────────┴─────────────────────────────────┘ ``` Google Analytics 4 is the first card in the **Analytics & Search** group. ### 2 — Click Connect and sign in as the right Google account The button changes to "Connecting…" and your browser leaves EOMA for Google's own sign-in page. Google will ask you to approve one thing: > **See and download your Google Analytics data** That is a read-only permission. There is no second checkbox and nothing to configure. Google shows this approval screen **every time**, even if you have approved EOMA before — that is deliberate on our side, not a sign that something went wrong. If Google is already signed in as the wrong account, use **Use another account** on that screen. Coming back and clicking Connect again will not offer you a choice; Google reuses whichever account the browser is already holding. When you approve, Google sends you back to Settings → Integrations. ### 3 — Pick your property A window opens by itself the moment you land back: ``` ┌─ Select Google Analytics properties ───────────────┐ │ [ Search properties or accounts ] │ │ Select all 2 of 12 selected │ │ ── ACME MEDIA ─────────────────────────── │ │ ☑ acme.com — GA4 ID: 312445901 │ │ ☐ acme.com — old UA copy ID: 118820043 │ │ ── CLIENT ACCOUNTS ────────────────────── │ │ ☐ shop.acme.com ID: 402118773 │ │ [ Cancel ] [ Save 2 … ] │ └────────────────────────────────────────────────────┘ ``` Properties are grouped under the Google Analytics **account** they sit in, because one Google login often reaches several. Tick the property that matches this brand's website and press Save. **Save stays greyed out until you tick at least one.** You can tick more than one — useful when a brand runs a main site and a shop as separate properties. Everything you tick is reported together. If the window did not appear, or you closed it, the card now shows a **Select properties** button. Same window, same result. ## What you'll see ### The card, before you connect The provider name and a **Connect** button. Nothing else. ### The card, after Google but before you pick A **Select properties** button and a small **Disconnect**. This is the state that looks finished and is not: the Google half is done, no property is attached, and the Web Analytics screen will still be empty. Finish step 3. ### The card, once a property is attached ``` Google Analytics 4 Connected · 2 properties Edit Disconnect ``` **Edit** reopens the same window with your current choices already ticked. ### "Owner only" Shown instead of any button when your role on this brand is not owner. Ask whoever created the brand to connect it. ### Where the data actually appears **SEO → Web Analytics** in the left sidebar. That screen is the only place GA4 data is shown. See [Web analytics](/docs/pages/web-analytics). EOMA reads your Analytics **live, each time that screen is opened**. Nothing is copied into EOMA overnight and there is no import to wait for — which also means the numbers you see there are Google's own, unchanged. ## Reading the numbers **"Connected · 2 properties".** Two GA4 properties are attached to this brand and both are being read. If this says one property and you run two websites, half your traffic is missing from EOMA — press Edit and tick the other. **"2 of 12 selected" in the picker.** Twelve properties are reachable by the Google account you signed in with; two are ticked. A large number here usually means you signed in with an agency or company account that also reaches clients' properties — check the account heading above each row before ticking, not just the property name. **"ID: 312445901".** Google's own number for that property, shown because two properties often carry near-identical names ("acme.com" and "acme.com — old"). If you are unsure which is live, this number is next to the property name in Google Analytics too. Match them. **"Data delayed ~15min" / "Data delayed ~24h" on Web Analytics.** Google itself does not publish the numbers instantly. Today's figures move for about a day after the fact, so a quiet-looking morning is often just Google catching up rather than a real drop. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No GA4 properties were found for this Google account." | You signed in with an account that cannot open any property. This is the single most common failure. | Disconnect, click Connect again, and use **Use another account** on Google's screen to pick the account that owns the property. | | The right property is missing from the list, but others are there | The account you used has access to some properties and not that one. | In Google Analytics, ask whoever administers the property to give your account access — Viewer is enough. Then Edit here. | | "Google Analytics connect failed: invalid state" | You left the Google screen open for more than 10 minutes before approving. | Click Connect and go through it again without pausing. | | "Google Analytics connect failed: access denied" | You pressed Cancel on Google's approval screen. | Click Connect and approve. | | The card says **Select properties** and Web Analytics is empty | Step 3 was never finished. The connection is half-made. | Press **Select properties** and pick one. | | Web Analytics says "Pick a Property" | The same half-made state, seen from the other screen. | Follow its button back to Integrations. | | Web Analytics says "Reconnect Required" | Your Google authorisation expired or somebody revoked EOMA's access from their Google account. | Disconnect, then Connect again. | | An email from EOMA saying an integration disconnected | The link broke and stayed broken. We send this **once** per disconnection, to the brand owner. | Reconnect. The alert clears itself once the connection works again. | | Web Analytics says "No Access to This Brand" | Your EOMA account is not on this brand's team. Reconnecting will not fix it. | Ask the brand owner to add you. | ## Related - [Web analytics](/docs/pages/web-analytics) — the screen this connection fills - [Google Search Console](/docs/integrations/search-console) — the other half of the same screen - [Integrations](/docs/settings/integrations) — every connection in one place --- # Google Search Console > Connect the Google account that verified your site, then choose which site to report on — the site picker does not open by itself. Source: https://eoma.ai/docs/integrations/search-console Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-07 --- ## In 10 seconds - Connecting is **two steps**: sign in with Google, then click **Select Site**. Unlike Google Analytics, the site list does *not* open on its own. You can tick **up to five properties**, and press **Edit** to change them later. - Your site must already be verified in Google Search Console. EOMA can only offer sites that Google already confirms you own. - Search Console reports in whole days and runs two to three days behind. It is never a "what happened this morning" tool. ## Before you start - You must be the **owner** of this brand in EOMA. Everyone else sees "Owner only". See [Integrations](/docs/settings/integrations). - Your website must be **verified in Google Search Console** already — that is Google's own check that you control the domain, done at `search.google.com/search-console`. EOMA cannot verify a site for you and cannot add one. - You need the Google account that holds that verification. It is frequently the developer's or the agency's account, not the one you read email with. - **Starter or higher.** Search Console is not part of the Free plan. On Free its card shows a lock and **Upgrade your plan** instead of Connect. See [Plans and limits](/docs/start-here/plans-and-limits). ## Steps ### 1 — Open Settings → Integrations ``` ┌─ Settings ───────────────────┬─────────────────────────────────────┐ │ Plan & Usage │ Analytics & Search │ │ Brand Profile │ ▮ Google Analytics 4 │ │ ● Integrations ← here │ ▮ Google Search Console [Connect] │ │ │ ▮ Google Drive │ └──────────────────────────────┴─────────────────────────────────────┘ ``` Second card in the **Analytics & Search** group. ### 2 — Click Connect and sign in Your browser leaves EOMA for Google. Google asks you to approve one thing: > **View Search Console data for your verified sites** Read-only, and it is the only permission requested. As with every Google connection here, the approval screen appears each time even if you have approved before. If the browser is already signed in to the wrong Google account, use **Use another account** on that screen. This is the failure that costs people the most time: the connection completes happily against an account that has no verified sites, and the next step then shows an empty list. Google returns you to Settings → Integrations. ### 3 — Click Select Site **Nothing opens by itself.** The card now reads: ``` Google Search Console [ Select Site ] Disconnect ↑ you must click this ``` Pressing it opens a list of the sites the account you signed in with has verified: ``` ┌─ Select Search Console properties ─────────────────────────┐ │ ☑ https://www.acme.com/ ◉ primary siteOwner │ │ ☑ sc-domain:acme.com ○ siteOwner │ │ ☐ https://shop.acme.com/ ○ siteFullUser │ │ [ Cancel ] [ Save 2 ] │ └────────────────────────────────────────────────────────────┘ ``` Tick every property that belongs to this brand — **up to five** — then press Save. EOMA checks with Google that each one really is verified for this account before saving, so an invalid choice is refused here rather than silently breaking your reports a week later. If any one of them fails that check, nothing is saved. **Primary** is the property your dashboard opens on, and the one EOMA's weekly SEO analysis reasons about. The others are collected too and you can switch to any of them from the Search Console page. **Why you might want more than one.** `https://www.acme.com/` and `sc-domain:acme.com` are different entries in Google covering different things: a URL-prefix property is blind to every other host you own, so if your blog lives at `blog.acme.com` none of its searches appear in that property at all. Selecting both is how you see all of it. **Changing your mind costs nothing.** Press **Edit** on the connected card and re-pick. Adding a property also makes EOMA go back and claim that property's history from Google, which it keeps for 16 months. ## What you'll see ### The card, before you connect The provider name and a **Connect** button. ### The card, after Google but before you choose A **Select Site** button and a **Disconnect**. This state looks connected and reports nothing. Finish step 3. ### The card, once a site is chosen ``` Google Search Console https://www.acme.com/ Edit Disconnect ``` The site address itself is the label, so you can confirm at a glance that you picked the right one. With several selected, the label names the **primary** one and counts the rest: ``` Google Search Console sc-domain:acme.com · +2 more Edit Disconnect ``` **Edit** reopens the picker with your current choice already ticked, so you can add, remove or re-assign primary without disconnecting. ### "Owner only" Shown instead of any button when your role on this brand is not owner. ### Where the data actually appears **SEO → Web Analytics** in the left sidebar, in the Search Console section below the traffic charts. See [Web analytics](/docs/pages/web-analytics). Search Console is a **separate connection from Google Analytics** and does not depend on it. If you connect only Search Console, that section still fills in — the Google Analytics part of the screen will show its own "connect" message above it, and that is expected rather than an error. EOMA reads Search Console live each time the screen is opened. Nothing is stored. ## Reading the numbers **The access level beside each site — for example `siteOwner`.** Google's own word for what your account may do with that site. Anything listed here can be read by EOMA; `siteOwner` means you verified it, the other levels mean somebody granted you access. It is shown so that, when two similar entries appear, you can tell the one you control from the one you were merely invited to. **Two entries for what looks like the same website.** `https://www.acme.com/` covers exactly that address and prefix. `sc-domain:acme.com` covers the whole domain including `shop.`, `blog.` and both `http` and `https`. If you have the domain entry, pick it — it is the more complete picture. Picking the narrow one is not an error, it just under-reports anything on another subdomain. **Why Search Console figures on Web Analytics stop short of today.** Google publishes this data in whole days and lands two to three days behind. Ask the screen for the last 24 or 48 hours and the Search Console section quietly reports the last 7 days instead and says so — because a shorter window there would always be empty, not because something failed. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "No verified sites found for the connected Google account." | The Google account you signed in with has not verified any site. The most common cause by far. | Disconnect, click Connect, and use **Use another account** to choose the account that verified the site. If no account has, verify the site at `search.google.com/search-console` first — that is a job on Google's side, not in EOMA. | | Your site is missing but others are listed | You are on the right account family but not the one holding this site's verification, or it was never verified. | Ask whoever verified the site to add your Google account as a user in Search Console, then reconnect. | | "That site is not verified for the connected Google account" | You picked a site the signed-in account cannot actually read. EOMA refused to save it. | Pick a different entry, or fix the access in Search Console first. | | "The site selection was not saved. Please try again." | The save did not take effect. It is not saved — the message is honest. | Press **Select Site** and choose again. If it repeats, contact support. | | "You can select up to 5 properties" | Five is the limit for one brand. | Untick one you no longer need, then add the new one. | | "Search Console connect failed: invalid state" | More than 10 minutes passed on Google's screen before you approved. | Connect again without pausing. | | The card still says **Select Site** and the section is empty | Step 3 was never finished. | Press **Select Site**. | | Web Analytics says "Pick a Search Console site" | The same half-made state seen from the other screen. | Follow its button back here. | | Web Analytics says "Reconnect Search Console" | Google rejected the saved authorisation — it expired or was revoked. | Disconnect, then Connect again. | | An email from EOMA saying an integration disconnected | The link broke and stayed broken. Sent once per disconnection, to the brand owner. | Reconnect; the alert clears when the connection works again. | ## Related - [Web analytics](/docs/pages/web-analytics) — the screen this connection fills - [Google Analytics 4](/docs/integrations/google-analytics) — the other half of the same screen - [Your brand on Google](/docs/pages/your-brand-on-google) — what Google's AI answers say about you --- # Google Drive > A one-click Google connection that lets EOMA create documents in your Drive — and cannot open anything already in it. Source: https://eoma.ai/docs/integrations/google-drive Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - One step, one click. There is no folder to choose and no second window, unlike the other two Google connections. - EOMA asks for the narrowest Drive permission Google offers: it can only touch **files it creates itself**. Your existing documents stay invisible to it. - Today this connection is groundwork. No screen in EOMA writes to your Drive yet — connecting it changes nothing you can see. Skip it until you need it. ## Before you start - You must be the **owner** of this brand in EOMA. Everyone else sees "Owner only". See [Integrations](/docs/settings/integrations). - Any Google account works. Unlike Google Analytics and Search Console, there is no property or verified site to match, so the "wrong account" problem barely applies here — pick the account whose Drive you want documents to land in. - No plan requirement. Integrations are available on every plan. ## Steps ### 1 — Open Settings → Integrations ``` ┌─ Settings ───────────────────┬─────────────────────────────────┐ │ Plan & Usage │ Analytics & Search │ │ Brand Profile │ ▮ Google Analytics 4 │ │ ● Integrations ← here │ ▮ Google Search Console │ │ │ ▮ Google Drive [Connect] │ └──────────────────────────────┴─────────────────────────────────┘ ``` Third card in the **Analytics & Search** group. ### 2 — Click Connect and approve Your browser leaves EOMA for Google's approval screen, which asks for one thing: > **See, edit, create and delete only the specific Google Drive files you use > with this app** Read that wording carefully, because it is the whole point of this connection. "Only the specific files you use with this app" means files EOMA creates. A document already sitting in your Drive is not one of them, and EOMA cannot list it, open it or change it. There is no broader Drive permission behind this one. Approve, and Google returns you to Settings → Integrations. ### 3 — Confirm the card That is the entire flow. There is no folder picker and nothing to select. ``` Google Drive Connected Disconnect ``` If it says anything else, see the table at the bottom. ## What you'll see ### The card, before you connect The provider name and a **Connect** button. ### The card, once connected The word **Connected** and a **Disconnect**. There is no name, count or folder beside it, because this connection has nothing to point at — it is permission to write into a Drive, not a link to one particular place in it. ### "Owner only" Shown instead of any button when your role on this brand is not owner. ### What changes elsewhere in EOMA Almost nothing, and this is worth being straight about. Connecting Drive today has one visible effect: when EOMA's agent works on your brand, "Google Drive" appears in its list of connected services, so it knows the option exists. **No screen in EOMA currently writes a document to your Drive.** The ability to create one is built and tested on our side, and nothing in the product calls it yet. If you connect Drive expecting reports or drafts to start appearing in it, they will not. Connect it when a feature asks you to; there is no benefit to connecting it early, and no harm in disconnecting it. For taking work out of EOMA today, use the download and copy options on an action's result instead. See [Actions](/docs/pages/actions). ## Reading the numbers **There is no number on this card, and that is the honest state.** Every other connection here counts something — properties, a site, a repository — because something in EOMA reads it. Drive has nothing to count yet. **"Connected" means only that the permission is stored.** It does not mean anything has been written, or ever will be until a feature uses it. Treat the word as "approved", not as "working". **One email per disconnection.** If the Google authorisation later breaks, EOMA emails the brand owner once — not once per attempt. Reconnecting resets it, so a future break notifies you again. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Google Drive connect failed: access denied" | You pressed Cancel on Google's approval screen. | Click Connect and approve. | | "Google Drive connect failed: invalid state" | More than 10 minutes passed on Google's screen before approving. | Click Connect and go through it without pausing. | | The card still says **Connect** after you approved | The approval came back but was not stored. | Click Connect once more. If it repeats, contact support. | | Nothing appears in your Google Drive | Expected. No feature writes to Drive yet. | Nothing to fix. Use the export options on an action's result instead. | | A message that the connection errored | The saved Google authorisation stopped working — usually revoked from the Google account's own security settings. | Disconnect, then Connect again. | | An email from EOMA saying an integration disconnected | The link broke and stayed broken. Sent once per disconnection, to the brand owner. | Reconnect, or disconnect the card if you do not need it. | | You want EOMA to stop having access entirely | Disconnecting here removes EOMA's stored permission. | Press **Disconnect**. To be doubly sure, also remove EOMA from your Google account's third-party access list. | ## Related - [Integrations](/docs/settings/integrations) — every connection in one place - [Actions](/docs/pages/actions) — where finished work is downloaded or copied today - [Google Analytics 4](/docs/integrations/google-analytics) — the Google connection that does feed a screen --- # WordPress > Connect your WordPress site so EOMA can publish the pages, posts and files it writes for you, instead of handing you a file to paste. Source: https://eoma.ai/docs/integrations/wordpress Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-24 --- ## In 10 seconds - A site you host yourself takes **three steps**: an Application Password, the EOMA plugin, then the three fields. WordPress.com takes one — you log in and you are done. - **The plugin is not optional.** Without it, work we publish to an Elementor site reports success and nothing a visitor sees changes. - WordPress publishes **live**. When you approve, the page is on your site right away. ## Before you start - You must be the **owner** of the brand in EOMA. Anyone else sees an "Owner only" label instead of the Connect button. See [Integrations](/docs/settings/integrations). - You need a WordPress login with the **Editor** or **Administrator** role. A Contributor or Author account cannot publish, and the connection will be refused. - **To install the plugin you need Administrator specifically.** WordPress only gives Administrators the right to install and activate plugins — an Editor can complete the connection and then find they cannot do step 2. If your own account is an Editor, ask whoever administers the site to do that step. - WordPress must be reachable from the internet and its REST API must not be blocked. Security plugins sometimes block it — the errors below name that case. ## Steps ### 1 — Open Settings → Integrations Settings is at the bottom of the left sidebar. Inside it, the **Workspace** group holds **Integrations**. ``` ┌─ Settings ─────────────────┬──────────────────────────────────┐ │ WORKSPACE │ Publishing — CMS & Sites │ │ Plan & Usage │ WordPress [ Connect ] ← │ │ Brand Profile │ GitHub [ Connect ] │ │ ● Integrations ← here │ Vercel [ Connect ] │ └────────────────────────────┴──────────────────────────────────┘ ``` Press **Connect** on the WordPress row. ### 2 — Say which kind of WordPress you have A window called **Connect WordPress** asks you to pick one: - **WordPress.com (hosted)** — you signed up at wordpress.com and your dashboard lives there. Addresses look like `myblog.wordpress.com`, or your own domain pointed at WordPress.com hosting. - **Self-hosted WordPress** — WordPress was installed on your own hosting, by you or by a developer, and the dashboard is at `yoursite.com/wp-admin`. Most business sites are this one. If you are not sure, open your dashboard and look at the address. `wp-admin` on your own domain means self-hosted. ### 3a — WordPress.com: log in Press **Continue to WordPress.com**. You approve the connection on their site and land back in EOMA with the row connected. Nothing to copy, and no plugin. ### 3b — Self-hosted: create an Application Password An Application Password is a separate password WordPress generates for one program to use. It is not your login password, and revoking it does not affect your login. 1. Log in at `yoursite.com/wp-admin` 2. Go to **Users → Profile** 3. Scroll to **Application Passwords**, type the name `EOMA`, press **Add Application Password** 4. Copy the 24-character password it shows, spaces and all WordPress shows that password **once**. If you close the screen before copying it, generate a new one — there is no way to read the old one back. ### 3c — Self-hosted: install the EOMA plugin Still in the EOMA window, under **2. Install the EOMA plugin**: 1. Press **Download the plugin (.zip)** 2. In WordPress admin go to **Plugins → Add Plugin → Upload Plugin**, choose the `.zip` you downloaded — do not unzip it — and press **Install Now** 3. When the upload finishes press **Activate Plugin** ``` ┌─ WordPress admin ────────────────────────────────────┐ │ Plugins ▸ Add Plugin ← the submenu │ │ [ Upload Plugin ] ← then this button │ └──────────────────────────────────────────────────────┘ ``` Two things go wrong here often enough to name: - **Skipping "Upload Plugin".** Without it WordPress only searches its own plugin directory, and this plugin is not in it — you will search for "EOMA" and find nothing. - **Installing and not activating.** Installing copies the files; activating is what turns them on. EOMA cannot see an installed-but-inactive plugin. If an older version is already on the site, choose **Replace current with uploaded** rather than deleting the old one first. Deleting the plugin from WordPress removes two things with it: the history EOMA uses to undo its own changes, and the stored content of your `/llms.txt`. Deactivating the plugin keeps both, so a deactivate-and-reactivate is safe when you are debugging. ### 3d — Self-hosted: enter the two things you collected Fill **Site URL**, **WordPress username** and **Application Password**, then press **Connect**. EOMA checks the credentials against your site before saving them. If the plugin is on the site, the window closes and the row turns connected. If it is not, the window stays open on **The plugin is not installed yet** — go back to step 3c, then press **I've installed it, check again**. ## What you'll see ### "Publishing — CMS & Sites" The group heading in Settings → Integrations, described as *"Where the agent can write pages, posts, schemas and root files."* WordPress is the first row in it. ### The WordPress row and its "Connect" button Icon, the name **WordPress**, and a **Connect** button on the right. If you are not the brand owner the button is replaced by **Owner only**. ### "Connect WordPress" — the two choices The window described in step 2. Each choice explains itself in two lines, so you can pick without knowing the word "hosting". ### "1. Create an Application Password in WordPress" Self-hosted only. The first of three numbered sections, opening with why the password is safe to give: *"This is NOT your WordPress login password… You can revoke it from WordPress in one click at any time."* Then the four steps from 3b. ### "2. Install the EOMA plugin" The second section, and the only one with a warning box around its explanation. It says the plugin is required whether or not you use Elementor, then gives the download link and the three install steps. ### "Video for this step" Under each of the first two sections. A short recording of that step being done in a real WordPress admin, with no sound — the instructions are written into the picture. **It only appears when EOMA is set to Turkish**, because the WordPress admin in the recording is Turkish; showing it to an English speaker would teach them labels their own screen does not have. The written steps are the same in both languages. ### "3. Enter the two things you just collected" The last section: Site URL, WordPress username and Application Password. A counter under the password field tells you whether what you pasted is the right length. ### "The plugin is not installed yet" The window you land on if the connection works but EOMA cannot find the plugin. It keeps the three install steps and the video, adds **I've installed it, check again**, and offers **I'll do it later** if you need to fetch an administrator. Leaving here is allowed — the site is connected either way — but publishing to it will not do what you expect until the plugin is on. ### The row after it connects The Connect button is replaced by your site address and a **Disconnect** button. That is the whole confirmation — there is no separate success screen. ## Reading the numbers **The counter under the password field — "18/24 chars".** An Application Password is exactly 24 characters, and EOMA counts what you pasted with the spaces removed. Anything under 20 turns orange and adds *"Application Passwords are 24 characters. This looks too short"* — which nearly always means you pasted your normal login password. At 24 it turns green and says *"looks right"*. Fix it here rather than pressing Connect and reading an error. **The site address shown on the connected row.** The address EOMA will write to. Check it is the one your visitors use. A staging address here means finished work lands on a site nobody reads. **What EOMA can send to WordPress.** More than to any other destination: | What EOMA writes | Lands as | Needs the plugin | |---|---|---| | A blog post | A post on your blog | Only if the page is built with Elementor | | A comparison page, a competitor-answer page, a rewrite of an existing page | A page | Only if the page is built with Elementor | | Your `llms.txt` file | A file at the root of your site | Yes, always | | FAQ or organization structured data | A snippet in the page head | Yes, always | **"Goes live on your site right away".** The line you will see later, when you approve a piece of work and choose WordPress as its destination. Take it literally: WordPress is a live destination. The moment you confirm, visitors see the page, and the action closes as done. The [GitHub](/docs/integrations/github) route is the one that gives you a review step first. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Owner only" instead of a Connect button | You are not the owner of this brand in EOMA. | Ask the brand owner to connect it. It is a one-time job. | | "That looks like a regular login password…" | You pasted your WordPress login password. | Go back to Users → Profile → Application Passwords and generate one there. | | "WordPress doesn't accept email here…" | You typed your email in the username field. | Use the username you type when logging in, not the email address. | | "Username or Application Password rejected by WordPress." | One of the two is wrong. The username is case-sensitive. | Re-check the username, then generate a fresh Application Password and paste the new one. | | "Application Passwords are disabled on this WordPress site." | The feature is switched off, usually by a security plugin. | Ask whoever maintains the site to turn it back on. The message in EOMA carries the exact line they need. | | "Your password is fine — your server is dropping the login header…" | Your hosting strips the Authorization header before WordPress sees it. Nothing you typed is wrong. | Send your host the `.htaccess` line the message shows. Re-generating the password will not help. | | "WordPress REST API isn't reachable." | A security plugin such as Wordfence is blocking `/wp-json`. | Ask your site maintainer to allow it, then press Connect again. | | "This WordPress user can't publish posts." | The account is a Contributor or Author. | Use an Editor or Administrator account. | | "We couldn't reach that site." or "The site took too long to respond (>10s)." | The address is wrong, or the site is down or very slow. | Open the address in a browser. If it loads, try again in a minute. | | "The plugin is not installed yet" after you installed it | Almost always: uploaded but never activated. | Open Plugins in WordPress. If the "EOMA Companion" row still says **Activate**, press it, then press **I've installed it, check again**. | | You cannot find **Add Plugin** in WordPress | Your WordPress account is an Editor. Only Administrators may install plugins. | Ask whoever administers the site to do step 3c. The connection itself is already done. | | The row is connected, work publishes, and the page does not change | The site uses Elementor and the plugin is missing. Elementor stores the page body separately, so a write can succeed and stay invisible. | Install the plugin (step 3c). Then re-publish the action. | | Everything connects, but nothing is ever published | Connecting gives EOMA somewhere to publish. It does not publish anything on its own. | Approve a piece of work on [Actions](/docs/pages/actions) and pick WordPress as the destination. | ## Related - [Actions](/docs/pages/actions) — where you approve the work that lands here - [GitHub](/docs/integrations/github) — the destination that opens a review instead of publishing live - [Integrations](/docs/settings/integrations) — every connection in one place --- # GitHub > Connect the repository your website is built from. EOMA opens a pull request you review and merge — it never writes to your main branch. Source: https://eoma.ai/docs/integrations/github Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - EOMA installs an app on the repositories **you choose**, and works only inside those. - Every change arrives as a pull request on its own branch named `eoma/draft/…`. Your main branch is never written to. - Nothing reaches your visitors until you merge. That step stays with you, always. ## Before you start - Your website must be built from a Git repository on GitHub. If your site is WordPress, connect that instead — see [Integrations](/docs/settings/integrations). - You need permission to install an app on the repository. On a personal account that is you; on an organisation it is an owner, or someone an owner has approved. - You must be the **owner** of the brand in EOMA, otherwise the row shows "Owner only". ## Steps ### 1 — Press Connect and choose the repositories Settings → Integrations → the **Publishing — CMS & Sites** group → the **GitHub** row. ``` ┌─ Settings ─────────────────┬──────────────────────────────────┐ │ WORKSPACE │ Publishing — CMS & Sites │ │ Plan & Usage │ WordPress [ Connect ] │ │ ● Integrations ← here │ GitHub [ Connect ] ← │ └────────────────────────────┴──────────────────────────────────┘ ``` Connect sends you to GitHub's own install screen. GitHub asks which account or organisation to install into, and then whether to grant **all repositories** or **only select repositories**. Choose *only select repositories* and pick the one your site is built from. This is the single most important decision on this page: it is the boundary of everything that follows. ### 2 — Choose the repository and the branch Back in EOMA the row shows **Select repo**. Pressing it opens **Select Repository** — the list of repositories you granted access to, and a **Branch** dropdown underneath. The branch you pick is the **base**: the branch EOMA compares against and points its pull requests at. For most sites that is `main`, which is already selected for you. Press **Save**. ### 3 — Approve a piece of work and choose GitHub Connecting gives EOMA somewhere to go. It does not send anything by itself. The first pull request appears when you approve a finished draft on [Actions](/docs/pages/actions) and pick GitHub as the destination. ## What you'll see ### The GitHub row and its "Connect" button In **Publishing — CMS & Sites**, second row. Non-owners see **Owner only** where the button is. ### GitHub's own install screen Not part of EOMA. This is where repository access is granted, and it is the only place it can be changed later. ### "Select Repository" — the repo list and "Branch" One line per repository you granted, showing its full name and its default branch. Below the selection, a **Branch** dropdown. At the bottom left, a link reading *"Don't see your repo? Manage repository access on GitHub →"*. ### The row once a repository is chosen The button is replaced by `owner/repo · branch`, with **Change repo** and **Disconnect** beside it. Change repo re-opens the same picker — you can move to a different repository without disconnecting and starting over. ### The pull request, on GitHub Not in EOMA. A branch called `eoma/draft/`, the changed files, and a pull request pointed at your base branch. EOMA also posts the link back onto the action, so you can open it from either side. ## Reading the numbers **What EOMA writes, exactly.** This is the blast radius, and it is small on purpose: | EOMA does | EOMA does not | |---|---| | Commit to a branch named `eoma/draft/`, one branch per piece of work | Commit to `main`, or to whatever base branch you selected | | Write the one file that piece of work produces | Touch any other file, or delete anything | | Open a pull request from that branch into your base | Merge it, approve it, or close it | | Read the repository to find the right path and the current file | Read repositories you did not grant access to | | Reset **its own** draft branch back to your base when it redoes a piece of work | Force-push, rewrite history, or change any branch that is not `eoma/draft/…` | If EOMA cannot do something within those lines, it stops and tells you. It has no second route. **"Don't see your repo?"** The picker lists only what the app was granted. A repository missing from the list is not broken — it was outside the selection you made on GitHub. The link takes you to the page where you widen it. You do not need to disconnect first. **The branch list can be shorter than your real branch list.** EOMA reads up to 100 branches. Your default branch is always at the top and always selectable, so a repository with hundreds of branches still works; a rarely-used branch may just not appear. If you need one that is missing, tell support. **Where the file lands.** EOMA picks a path from the kind of work: a blog post goes to `content/blog/.md`, a page to `content/pages/.md`, your `llms.txt` to the root of the repository. Structured-data snippets go to `content/schema/`. If your repository is laid out differently, say so before the first publish and the paths can be set per brand — a pull request in the wrong folder is easy to close, but easier not to open. **"Opens a pull request in your repo — merging stays with you".** The line you will see when you approve. The confirmation that follows spells out the same three facts: EOMA opens the pull request and writes nothing directly, you see the change line by line, nothing goes live until you merge. **Disconnecting removes the installation.** Pressing **Disconnect** removes EOMA's app from your GitHub account as well as the connection inside EOMA. Branches and pull requests already opened stay where they are — they are yours. You can delete or merge them at your own pace. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Owner only" instead of a Connect button | You are not the owner of this brand in EOMA. | Ask the brand owner to connect it. | | "No repositories found." after connecting | The app was installed, but no repository was selected — or it went onto a different account than the one that owns your site. | Use *Manage repository access on GitHub* and add the repository. It appears in the picker straight away. | | An action says the publishing target is not configured | GitHub is connected, but no repository was ever chosen. | Settings → Integrations → GitHub → **Select repo**. | | A pull request opens against the wrong branch | The base branch chosen at connect time is not the one your site builds from. | **Change repo**, pick the right branch, and save. The next pull request uses it. | | The pull request is correct, but nothing changes on your site | It has not been merged. EOMA never merges. | Merge it on GitHub. Your normal deploy runs as it always does. | | You merged, and the action still shows as in progress | EOMA is not told when a pull request is merged — nothing watches for it yet. | Say so in support and the action gets closed. It is a bookkeeping gap on our side; your change is live regardless. | | The repository is on an organisation and the install needs approval | GitHub asks an organisation owner to approve new apps. | An owner approves it once from GitHub's settings, then continue at step 2. | ## Related - [Actions](/docs/pages/actions) — where you approve the work that becomes a pull request - [Vercel](/docs/integrations/vercel) — for a Vercel site that deploys from this same repository - [WordPress](/docs/integrations/wordpress) — the destination that publishes live instead --- # Vercel > Connect Vercel so EOMA can publish redirects at the CDN and find the repository the rest of your site is built from. Source: https://eoma.ai/docs/integrations/vercel Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-23 --- ## In 10 seconds - **Redirects** EOMA prepares for you in Vercel itself. They wait in your Vercel dashboard until **you** press Publish — EOMA never presses it. - **Everything else** — page text, titles and descriptions, structured data — cannot be changed from outside Vercel. That work goes to the repository your site is built from. - Connecting Vercel is also how EOMA finds out **which repository** that is, so you never have to type its name. ## Before you start - You must be the **owner** of the brand in EOMA. Others see "Owner only" instead of the Connect button. - You need to be able to install an integration on the Vercel account or team that owns your project. ## Steps ### 1 — Open Settings → Integrations Settings is at the bottom of the left sidebar; **Integrations** is in the **Workspace** group. Vercel is in **Publishing — CMS & Sites**. ### 2 — Install the EOMA integration Press **Connect**, then **Continue to Vercel**. Vercel shows you what EOMA is asking for and where to install it. There is nothing to copy and no token to create. EOMA asks to read your projects and their settings, and to work with deployments. It does not ask for your environment variables, your domains or your billing. ### 3 — Pick the project that is your website You come straight back to EOMA, and it lists the projects the installation can see. Each row shows how that project deploys: | Badge | What it means | |---|---| | Deploys from *owner/repo* | Your site is built from that repository. This is the normal case | | Direct deploys (CLI / v0) | The files were uploaded rather than built from a repository | Choose the one that serves your website and press **Save project**. ### 4 — Connect GitHub If your project deploys from a repository, EOMA now tells you which one and takes you to GitHub. When you install EOMA there and grant it access to that repository, **the publishing target is filled in for you** — no repository picker, no branch to choose. If the repository is not in your own account, EOMA names the account it is in. Forward that to whoever built your site: they need to install EOMA on GitHub and give it access to that repository. ## What you'll see **"One more step — connect GitHub".** The screen after you pick a project, when that project deploys from a repository. It names the repository and explains that changes will arrive there as pull requests you review and merge. **The repository already selected.** After the GitHub install, the publishing target is set. If you look at the GitHub row you will see the repository and branch already filled in. **A named account when access is missing.** If EOMA was not given access to the repository your Vercel project names, it says so and names the account that owns it, rather than showing you an empty list. ## Reading the numbers **The deploy badge is the whole decision, and it is the only one you have to read.** | Badge | What it means | What happens | |---|---|---| | Deploys from *owner/repo* | Your site is built from that repository | Redirects are prepared in Vercel for you to publish. Everything else goes to that repository as a pull request you review and merge | | Direct deploys (CLI / v0) | The files were uploaded, not built from a repository | Redirects still work the same way. There is no repository for the rest, so those actions come to you ready to apply | **What EOMA can publish through Vercel, and what it cannot.** Vercel sits in front of your site as a CDN, and it can be told to answer a request differently before your site ever sees it. That is enough for a **redirect** — an old address that should send visitors to a new one — so EOMA prepares those in Vercel directly. It is not enough for anything that lives inside the page. A title, a description or a block of structured data is part of the HTML your site returns, and the CDN cannot reach into that. Vercel is where your site is hosted, not where it is written: when your project is connected to a Git repository, that repository is the real source, and Vercel deliberately does not let anything change it from the outside — ask its API for the files of a Git-based deployment and it declines. That is a good thing. It means nobody, including us, can quietly rewrite your site behind your developer's back. So that work goes where your site is actually written. EOMA opens a pull request on your repository, you review it and merge it, and Vercel builds and publishes the result the way it always does. **Nothing goes live in Vercel until you press Publish.** When EOMA prepares a redirect, it lands in your project's staging list under **CDN → Routing Rules**, exactly where a change you made yourself would land. Production traffic is untouched until you publish it. EOMA holds a credential that *could* publish it — we do not, by choice, because it is your site. If you already have an unpublished routing change of your own waiting there, EOMA will stop and say so rather than adding to it. Publishing yours would otherwise publish ours in the same press, and the other way round. The reason to connect Vercel at all is that **the repository is often not yours to name.** It frequently belongs to the agency, freelancer or developer who built the site, under an account you have never seen. Vercel knows which one it is. Connect it, and we read that from your project instead of asking you a question you cannot answer. **What EOMA can read.** Your projects and their settings, and your deployments. Not your environment variables, not your domains, not your billing. You can remove the integration from your own Vercel dashboard at any time, and EOMA finds out immediately when you do. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Owner only" instead of a Connect button | You are not the owner of this brand in EOMA. | Ask the brand owner to connect it. | | "This installation can't see any projects." | The integration was installed on an account or team that does not hold your project, or it was not given access to it. | On Vercel, open the EOMA integration and give it access to the project that serves your site. | | "Install the EOMA integration on Vercel first." | The project list was requested before the install finished. | Press Connect again and complete the Vercel step. | | EOMA says the repository is in someone else's account | Your site is built from a repository you do not own — usually your developer's or agency's. | Send them the sentence EOMA shows. They install EOMA on GitHub and grant access to that repository. | | Your Vercel project deploys from GitLab or Bitbucket | EOMA can only open pull requests on GitHub today. | Page edits and structured data will come to you ready to apply instead of being published for you. | | "Vercel rejected our access." | The integration was removed or its permissions changed on the Vercel side. | Disconnect in EOMA and install it again. | | "You have an unpublished routing change waiting" | Your Vercel project has a staged rule nobody has published yet. | Open **CDN → Routing Rules** in Vercel and either publish or discard what is waiting, then approve the action again. | | A redirect was approved but the page still goes to the old address | It is prepared, not published. That press is yours. | Open **CDN → Routing Rules** in Vercel and select **Publish**. | ## Related - [GitHub](/docs/integrations/github) — where a Vercel site's changes are actually published - [Actions](/docs/pages/actions) — where you approve the work - [Integrations](/docs/settings/integrations) — every connection in one place --- # Plan and usage > What you are paying for, how much of this week's allowance you have left, and what changes the moment you upgrade. Source: https://eoma.ai/docs/settings/plan-and-usage Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-06 --- ## In 10 seconds - One bar shows the one thing this screen meters: how many done-for-you jobs EOMA is running for you right now, against the seats your plan holds. - Running out never breaks the product. It stops one specific thing, and the rest keeps working. - When you upgrade, the unused days of your current plan are deducted and the difference is charged to your card right then. The new plan opens the moment that payment goes through, on this same screen. ## Before you start - A brand must be selected. With no brand active the panel says "Select a brand to see plan and usage." - Payments go through Stripe, our payment provider. The free plan needs no card: nothing is charged on it. - Nothing else. This page works from your first day. ## Steps ### 1 — Open Settings, then Plan & Usage ``` ┌─ sidebar ──────┐ ┌─ Settings ───────────────┐ │ Dashboard │ │ WORKSPACE │ │ Actions │ │ ● Plan & Usage ← here │ │ Settings ← │ │ Brand Profile │ └────────────────┘ │ Integrations │ └──────────────────────────┘ ``` The plan chip at the bottom of the left sidebar — the one showing your tier name — is a shortcut to this same page. ### 2 — Read the bar before you do anything else It is the answer to "can I start another job right now". The bar turns amber at 80% used and red with a warning triangle when every seat is busy. ### 3 — Press "Upgrade plan" only when the bar is the reason Upgrading opens a panel of tier cards. Pick one, tick the agreement box, and you pay in Stripe's card form. You come back to this page with your new limits already applied. ## What you'll see ### The plan name The plan's name at the top (Free, Starter, Growth or Scale), with one sentence under it about where the plan stands. On the free plan that sentence reads "No card needed; nothing is charged on this plan." ### The three plan rows underneath it **Brand profile depth** — "Basic" or "Deep AI training". **Social listening** — "Disabled", "Weekly", "Daily" or "Once at onboarding". **Support** — a dash, "Email", or "Priority". ### "Upgrade plan" and "Cancel subscription" Two buttons in a row. Cancel appears only when there is a paid subscription to cancel; it is hidden once a cancellation is already scheduled. ### "Usage" — the bar One bar, "Done-for-you executions", showing used / limit: how many seats are taken right now against how many your plan holds. It carries no reset date — seats free as work finishes. Your tracked-prompt count is not metered on this screen; the prompt ceiling for each plan is on [Plans and limits](/docs/start-here/plans-and-limits). ### After a payment Paying does not bring you back here. The payment page says "Payment received" while your plan is being activated, and the dashboard opens by itself when it is ready. If activating takes longer than expected, the page says so: your payment was received, so do not pay again. ## Reading the numbers **Done-for-you executions — for example 2 / 3.** This is how many actions EOMA is carrying out *for* you right now, against how many it can hold at once. Growth gives you 3 seats, Scale 6, Enterprise a custom number. A seat is taken the moment you press Start on an action and frees the moment that action reaches done or you remove it — so "2 / 3" means two seats are taken and one is open. There is no weekly quota behind this bar and nothing resets on Monday: over a quiet month you can start as many actions as you like, three at a time. Starter is the exception: it gets 3 executions in total, which do not renew — once they are spent, EOMA does no further done-for-you work on the account. When every seat is busy you can still see every recommended action, read it, and do it yourself: the recommendation list is not metered on any tier. Only pressing Start takes a seat, and each action takes exactly one no matter how long it runs. **A red bar with a warning triangle.** Every seat is busy. On Growth and Scale it clears the moment one of the jobs in flight finishes or you remove it; on Starter the three executions do not renew, so it clears only by upgrading. **The tier cards inside "Upgrade plan".** Starter is $49 a month: 30 AI prompts, 200 keywords tracked, 3 done-for-you executions in total, no social listening, no support tier. Growth is $99 a month: 100 AI prompts, 2,000 keywords, 3 done-for-you actions at a time, email support. Scale is $249 a month: 200 AI prompts, 4,000 keywords, 6 done-for-you actions at a time, priority support. A Monthly | Yearly switch above the cards picks the billing period; a year costs ten months (two months free). Moving from monthly to yearly is charged right away, with the unused days of your month deducted; moving from yearly to monthly takes effect when the year you paid for ends. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Couldn't load plan" with a Try again button | The read failed. Your plan is unaffected. | Press Try again. If the message mentions your session, refresh the page to sign back in. | | "No active plan" | No plan has resolved for this brand. | Open Upgrade plan and choose one. If you believe you already pay, contact support before paying again. | | "Your payment was received; activating your plan is taking longer than expected." | Your card was charged and our record of it has not caught up. | **Do not pay again.** Press Check again; it usually settles within minutes. If it does not, contact support quoting the date. | | "Checkout did not complete. No card was charged." | You closed or cancelled the card form. | Nothing was taken. Start the upgrade again when ready. | | "Your cancellation is confirmed" | A cancellation is scheduled. | You keep full access until the date shown, then billing stops. Your data, brand profile and history stay intact. Restarting later is a normal upgrade — no setup fee. | | Enterprise has no card form | Enterprise is arranged by agreement, not self-serve. | Pressing its button records your interest and someone contacts you. | ## Related - [Plans and limits](/docs/start-here/plans-and-limits) — what each tier is for, before you compare numbers - [Actions](/docs/pages/actions) — where a done-for-you execution actually gets spent - [Privacy](/docs/settings/privacy) — the billing identity data held alongside your subscription --- # Brand profile > The six fields that decide which questions EOMA asks the AI engines on your behalf — and therefore what every other screen can measure. Source: https://eoma.ai/docs/settings/brand-profile Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - What you write here is turned into the questions EOMA asks ChatGPT, Google and the other engines about your market. - A vague description produces vague questions, and vague questions produce a score that is not about your business. - This is the one screen where twenty minutes of your own writing changes every other screen. ## Before you start - A brand must be selected. With no brand active the panel says "No active brand selected." - Have your website address to hand, and know which countries you actually sell in. - Nothing else is required. You can fill this in on day one and improve it later. ## Steps ### 1 — Open Settings, then Brand Profile ``` ┌─ Settings ───────────────┐ │ WORKSPACE │ │ Plan & Usage │ │ ● Brand Profile ← here │ │ Integrations │ └──────────────────────────┘ ``` ### 2 — Fill the fields top to bottom Every field with a small ⓘ next to its label opens a panel showing what that field is used for, what a good answer looks like, what a bad one looks like, and a worked example. Read the one on **Description** before you write yours. ### 3 — Press "Save Changes" The button sits top right and briefly changes to "Saved". Name and Description are required; leaving either empty blocks the save and marks the field in red. ### 4 — Press "Re-analyze" after a meaningful change Next to Save. It re-reads your website and knowledge base and rebuilds what EOMA understands about your business. A progress panel opens while it runs. Use it after you change your website address, rewrite the description, or add a substantial document — not after fixing a typo. ## What you'll see ### The header — "Brand Profile", with "Re-analyze" and "Save Changes" ### "Business Info" — Name and Description Name is the identity used everywhere in the product and in the questions asked about you. Description is a free-text box that grows as you type. ### "Website" One address field. Typing `example.com` and clicking away turns it into `https://example.com` for you. ### "What's your service area?" Four choices: Single location / neighborhood, City-wide, Country-wide, Global / multi-country. Choosing either of the first two opens a city box — type a city or district and press Enter to add it as a chip. ### "Market Locations" "Select the countries or regions where your business operates." Opens a picker where you choose countries by region or by search, and confirm with Apply. ### "Known Competitors" "Competitors you already know about — helps us find more like them." This list is **read-only here**. It shows the rivals currently tracked, with a line telling you how many, and points you to the Share of Voice card in AI Visibility to add or remove one. That is the only place editing happens. ### "Knowledge Base" Two panels: files you upload (PDF, DOCX, TXT, MD) and blocks of text you paste in with a title. Each item saves on its own the moment you add it — there is no separate save step for this section. ## Reading the numbers There are no metrics on this screen. There is one number, and one consequence chain that matters more than any number in the product. **"Tracking 19 competitors."** The count of rivals EOMA compares you against on every AI answer it collects. Nineteen is a healthy list for a normal market; a count of zero means Share of Voice has nothing to rank you inside, so that screen will look empty until the list fills. The list grows on its own as EOMA discovers rivals, so a low number on a new brand is normal rather than broken. **Why a thin profile produces weak measurement.** When EOMA builds the set of questions it will ask the AI engines, it hands the generator exactly five things from this screen: your **name**, your **description**, the **domain** taken from your website address, your **market locations** (which also decide which language each question is asked in), and your **competitors**. Nothing else about your business reaches that step. So: - A description reading "We do marketing" produces generic questions that any agency in the world would also match. Your brand then fails to appear in the answers, and the score is honest about a market you were never really measured in. - A description naming what you sell, to whom, and where — "a specialty coffee roaster in Austin sourcing single-origin beans for remote workers" — produces questions real customers actually ask, and a score you can act on. - An empty **website** field means EOMA has no domain to look for in the sources of an AI answer. Citation counts, Opportunity gaps and Your brand on Google all compare against that domain, so they have nothing to compare. - Empty **market locations** means no country and no language context. For a Turkish business that is the difference between being measured in Turkish, where your customers ask, and in English, where they do not. **The cap on how many questions get built** comes from your plan, not from this screen: 30 on Starter, 100 on Growth, 200 on Scale, 400 on Enterprise. A better profile does not buy more questions — it makes the ones you have worth asking. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Failed to load brand data. You can still edit and save." | The read failed; the form is empty rather than wrong. | Do **not** press Save on the empty form — that would overwrite good data with blanks. Refresh first. | | "Name is required." / "Description is required." | One of the two mandatory fields is blank. | Fill it. Nothing saves until both are present. | | "Failed to save. Please try again." | The write did not land. | Your typing is kept as a draft in this browser, so retry rather than starting over. | | "No competitors tracked yet." | Nothing has been discovered or added for this brand yet. | New brands start here. Add one on the Share of Voice card if you want to seed the list. | | The competitor box will not let you type | It is deliberately read-only. | Add and remove competitors on the Share of Voice card in AI Visibility. | | Re-analyze finishes but the profile text looks unchanged | Re-analyze rebuilds EOMA's internal understanding of your brand; it does not rewrite the words you typed. | That is expected. Your fields are yours; the analysis result feeds other screens. | ## Related - [Why prompts matter](/docs/concepts/why-prompts-matter) — how a description becomes a question asked to an AI engine - [Share of Voice](/docs/pages/share-of-voice) — where competitors are actually added and removed - [Your first week](/docs/start-here/your-first-week) — the order to do this in --- # Integrations > The one screen where you connect your analytics, your website and your publishing tools — and the map of which guide to open for each. Source: https://eoma.ai/docs/settings/integrations Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-07 --- ## In 10 seconds - Two groups: what EOMA reads from, and where it can publish for you. - Each row is one connection with one button. Connecting is always the same shape; only the credentials differ. - Every row works on every plan except Google Search Console, which starts at Starter. ## Before you start - **You must be the brand owner.** Everyone else sees "Owner only" instead of a button, on every row that does something — that lock is how the page tells you, and it appears on each row rather than once at the top. - Have the account you are connecting open in another tab. Several of these ask you to copy a token out of the other product. - **Your plan matters for one row only.** Google Analytics and the five publishing connections (WordPress, GitHub, Vercel, HubSpot, Wix) work on every plan, Free included. Google Search Console starts at Starter: on Free its card shows a lock and **Upgrade your plan** instead of Connect. - Nothing else. Connections can be made on your first day or your hundredth. ## Steps ### 1 — Open Settings, then Integrations ``` ┌─ Settings ───────────────┬────────────────────────┐ │ WORKSPACE │ Analytics & Search │ │ Plan & Usage │ Google Analytics 4 │ │ Brand Profile │ [ Connect ] │ │ ● Integrations ← here │ Google Search Console │ └──────────────────────────┴────────────────────────┘ ``` ### 2 — Find the row you want and press its button The button tells you which kind of connection it is. **Connect** starts a Google or GitHub sign-in in a new window. **Connect** on WordPress or Vercel opens a short panel that walks you through creating a token in that product first. ### 3 — Finish the second step where there is one Some connections are not done when the sign-in is. Google Analytics returns and asks you to pick properties. Search Console asks you to pick one or more properties. GitHub asks you to pick a repository and branch. Vercel asks you to pick a project. Until you do, the row keeps showing that request instead of a connected label. Google Analytics, Search Console and GitHub also let you **change** that choice later: the connected row carries an **Edit** button beside Disconnect, and it reopens the same picker with your current selection already ticked. ### 4 — Open the guide for the one you are connecting Each connection has its own page with the exact clicks, the permissions it asks for, and what EOMA does with the access. They are listed under Related and in the table below. ## What you'll see ### The line at the top "Connect your site and accounts; drafts you approve go live from here on their own." Below it the page goes straight into the groups — there is no introductory card. If your site is already connected, one panel sits above the groups: "Your site is connected", confirming that approved fixes publish themselves. If nothing is connected yet, that space is empty. ### "Analytics & Search" "Connect your data sources so EOMA can read your traffic, your search results and how your content performs." Two rows: **Google Analytics** · **Google Search Console**. A third row, **Google Drive**, appears only if your brand connected Drive before; it keeps its Disconnect button there. See [Google Analytics](/docs/integrations/google-analytics), [Search Console](/docs/integrations/search-console), [Google Drive](/docs/integrations/google-drive). ### "Publishing · CMS & sites" "Where EOMA can publish pages, posts and site files for you once you approve them." Five rows: **WordPress** · **GitHub** · **Vercel** · **HubSpot** · **Wix**. See [WordPress](/docs/integrations/wordpress), [GitHub](/docs/integrations/github), [Vercel](/docs/integrations/vercel). There is no social accounts group: EOMA does not publish to social profiles. It **reads** public social posts for Social Listening, which needs no account of yours, and it **writes to your own site** through the publishing rows above. ### The right-hand side of every row This is the whole state of the connection, in one place: | What you see | What it means | |---|---| | **Connect** | Not connected. Pressing it starts the flow. | | **Connecting…** | The sign-in window is open or the token is being checked. | | A name — a site address, a repo, a shop, a project | Connected, and this is which account. | | **Select properties / Select Site / Select repo / Select project** | Signed in, but you have not told EOMA which one to use. Not finished. | | **Connected · 3 properties** | Google Analytics, with three properties attached. **Edit** changes the set. | | `owner/repo · main` with **Change repo** | GitHub, pointed at that repository and branch. | | A lock and **Upgrade your plan** | Your plan does not include this connection. Today that is Google Search Console on Free. If you connected it before a downgrade, the card keeps its connected state and its Disconnect button instead. | | **Owner only** | You are not the brand owner. | | A red message with **Retry** | The last attempt failed. The message is what the other service said. | | **Disconnect** | Removes EOMA's access. | ## Reading the numbers Only one number appears on this screen. **"Connected · 3 properties" on Google Analytics.** Three GA4 properties are attached to this brand, and the traffic figures on Web analytics are the sum across all three. That matters more than it looks: attaching a property you do not own the traffic of will quietly inflate every number downstream, and attaching none means Web analytics has nothing to show even though the row says connected. Press **Edit** and check the list matches the sites this brand actually is. Everything else here is a state, not a measurement. The useful reading is the negative one: **a row that shows "Select properties", "Select Site", "Select repo" or "Select project" is not connected in any way that does work.** The sign-in succeeded and the choice never got made, so EOMA holds an access token pointing at nothing. It looks half-green and behaves like off. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "Owner only" on every row | You are a member of this brand, not its owner. | Ask the owner to connect it. The backend refuses the write either way. | | "Token expired — reconnect" | The other service invalidated the access. This is normal after a password change or a revoked app. | Press Connect again. Nothing else is lost. | | A picker opens with nothing in it | The account you authorised owns no site, property, project or repo that EOMA can see. | Sign in again with the account that owns it. For Search Console the site must be verified there first. | | Connect does nothing | A pop-up blocker stopped the sign-in window. | Allow pop-ups for eoma.ai and press it again. | | Connected, but the screen that needs it is still empty | Connecting grants access; the data is collected on the next scheduled run. | Give it a cycle before treating it as broken. | ## Related - [Google Analytics](/docs/integrations/google-analytics) — the most common first connection - [What EOMA does and what you do](/docs/start-here/what-eoma-does-and-what-you-do) — what access buys you --- # Account > Your sign-in email, your password, the language EOMA operates in, and the optional second step at login. Source: https://eoma.ai/docs/settings/account Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Four blocks, all about you rather than your brand: email, password, language, two-factor sign-in. - Changing your email does not take effect until you click the link sent to the new address. - Changing the language changes the language EOMA writes your content in, not only the buttons. ## Before you start - You are signed in. That is the whole requirement. - To change your password you need your current one, even though you are already signed in. - To turn on two-factor sign-in you need an authenticator app on your phone. ## Steps ### 1 — Open Settings, then Account ``` ┌─ Settings ───────────────┐ │ PREFERENCES │ │ ● Account ← here │ │ Notifications │ │ Privacy │ └──────────────────────────┘ ``` The language shortcut in your account menu at the bottom of the sidebar also lands here. ### 2 — Change your email, if you need to Type the new address twice and press **Send confirmation link**. A link goes to the **new** address. Keep signing in with the old one until you click it. ### 3 — Change your password, if you need to Current password, new password, new password again, then **Update password**. Minimum six characters. ### 4 — Change the language only deliberately Pick a language chip and a warning appears with a confirm button. Read it before pressing. ## What you'll see ### The header — "Account" "Manage your account email. Changing it sends a confirmation link to the new address — the change takes effect once you click it." ### "Current email" One row showing the address you sign in with today. ### The email form — "New email" and "Confirm new email" Two fields and **Send confirmation link**. Once sent, the form is replaced by "Confirmation sent." and the address it went to. ### "Password" "Change the password you use to sign in. You'll need your current password to confirm." Three fields — Current password, New password, Confirm new password — with "Min 6 characters" under the middle one. On success the form is replaced by "Password updated." ### "Language" "The language EOMA operates in — its interface and the AI content generated for you." A row of chips, one per supported language, with the current one filled in. Selecting a different one shows a warning panel headed "Switching language is a heavy change", with **Yes, switch to …** and **Keep current language**. ### "Two-factor authentication" "Add a second step at sign-in using an authenticator app (Google Authenticator, Authy, 1Password)." Shows **On** or **Off**. Pressing **Enable 2FA** shows a QR code to scan and asks for the six-digit code from your app. After that it shows a set of recovery codes **once** — each works a single time and lets you in if you lose your phone. Once on, the block offers **Regenerate recovery codes** and **Disable 2FA**. ## Reading the numbers Three numbers appear on this screen, and each one is a rule rather than a measurement. **"Min 6 characters."** The shortest password EOMA will accept. It is a floor, not a recommendation — six characters is guessable by a machine in moments. If this account can start work that publishes to your live website, treat it like the login to that website and use a long passphrase or a password manager. **The six-digit code.** The number your authenticator app shows, which changes every thirty seconds. It is proof you have your phone, not a second password. EOMA never sends it to you; if something asks you for it by email or message, that is not us. **"8 recovery codes remaining."** How many one-time codes you have left to sign in with if your phone is lost or replaced. Each sign-in that uses one spends it permanently. At zero, and with no phone, nobody can let you back in from this screen — so store them somewhere that is not the phone they replace, and regenerate the set when the count gets low. Regenerating cancels every old code immediately. ## When it goes wrong | You see | Why | What to do | |---|---|---| | "The two email addresses don't match." | The two fields differ. | Retype the second one. Copy-paste is safer than typing twice. | | "That's already your current email." | The new address is the one you already use. | Nothing to change. | | "Too many requests. Please wait a minute before trying again." | Several confirmation emails were requested in quick succession. | Wait a minute. Check spam before asking for another. | | The confirmation email never arrives | It goes to the **new** address, not the current one. | Check that inbox and its spam folder. Your old address keeps working until you click the link, so nothing is locked. | | "Your current password is incorrect." | The current-password field is wrong. | If you cannot remember it, sign out and use the password reset link on the sign-in page. | | "Your new password must be different from your current one." | The old and new password are identical. | Choose a different one. | | The language chips are greyed out | No brand is selected. | Pick a brand in the sidebar first — language is stored on the brand. | | Your content is a mix of two languages after switching | Expected for a while. EOMA re-generates AI content in the new language over the following cycles rather than all at once. | Wait for the next cycles. Nothing is lost. | | Enabling 2FA returns "Something went wrong. Please try again." | The enrolment did not complete. | Try once more, then contact support before repeating it further — a half-finished enrolment is worth checking rather than retrying blindly. | ## Related - [Privacy](/docs/settings/privacy) — the personal data held for billing, and how to delete it - [Notifications](/docs/settings/notifications) — what EOMA sends you - [Plan and usage](/docs/settings/plan-and-usage) — the brand-level settings, as opposed to these personal ones --- # Notifications > What EOMA sends you today, and why there is nothing to configure here yet. Source: https://eoma.ai/docs/settings/notifications Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - There are no notification settings yet. The page says so, and it is honest. - EOMA already notifies you about four things, and all four are about actions. - The one thing you *can* switch off — marketing email and SMS — lives on the Privacy page, not this one. ## Before you start - Nothing. This page works from your first day and needs no setup. ## Steps ### 1 — Open Settings, then Notifications ``` ┌─ Settings ───────────────┐ │ PREFERENCES │ │ Account │ │ ● Notifications ← here │ │ Privacy │ └──────────────────────────┘ ``` ### 2 — Read the one line, and leave "Notification settings coming soon." Nothing on this screen turns notifications on or off. If you came here to stop an email, the section you want is [Privacy](/docs/settings/privacy) for marketing messages — action emails cannot currently be switched off. ## What you'll see ### "Notifications" and "Notification settings coming soon." A heading and a single line of muted text. That is the whole notifications part of the page today. ### What EOMA sends you, even though you cannot configure it here Four moments in the life of an action, sent both as an in-app notification on the bell and as an email to your account address: - a **draft is ready** for you to look at - EOMA **needs a detail from you** before it can continue - EOMA has **finished** the work - EOMA **could not finish** and has stopped A fifth email is sent when EOMA opens a pull request on your website's repository. ## Reading the numbers There are no metrics on this screen. Two numbers govern what lands in your inbox, and both are worth knowing before you decide whether something is broken. **Four states, not four emails per action.** An action sends you a message when it crosses one of the four moments above, once each. A single piece of work that goes draft-ready, then needs a detail, then finishes will mail you three times over several days. Ten open actions moving at once is what a busy week looks like — it is not a loop or a duplicate. **A fifteen-minute check.** EOMA looks for actions that have crossed one of those moments every fifteen minutes and sends what it finds. So the gap between something becoming true in the product and the email arriving is up to a quarter of an hour, never longer under normal conditions. If the bell shows a draft is ready and no email has arrived after twenty minutes, that is worth reporting; five minutes is not. **Zero configurable settings.** The count of things you can change on this screen. Everything above happens whether you want it or not, which is the reason this page carries a "coming soon" line rather than an empty list. ## When it goes wrong | You see | Why | What to do | |---|---|---| | You want to stop the action emails | There is no switch for them yet. | Contact support and say so — the volume of that request is what moves it up the list. | | You want to stop campaign and announcement emails | Those are separate, and they *do* have a switch. | Open [Privacy](/docs/settings/privacy) → Commercial Message Preferences and untick email or SMS. | | The bell shows something the email never mentioned | The two surfaces cover the same four states, but an email can be caught by a spam filter. | Check spam, then add our sending address to your contacts. | | Emails arrive at the wrong address | Action emails go to your account email. | Change it on [Account](/docs/settings/account) — remember it only switches after you click the confirmation link. | ## Related - [Privacy](/docs/settings/privacy) — the marketing-message switches, and your billing data - [Actions](/docs/pages/actions) — the four moments these notifications are about - [Account](/docs/settings/account) — which address the emails go to --- # Privacy > The personal data EOMA holds for billing, the marketing messages you can switch off, and the two different ways to have data deleted. Source: https://eoma.ai/docs/settings/privacy Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - This screen covers **billing identity only** — your identity number, your phone, and the date you consented. - Withdrawing that consent deletes those three things and cancels a paid subscription with them. - Deleting your whole account is a different, larger request, made by email. ## Before you start - A brand must be selected. The data shown is held against a brand, not against you personally. - Know which of the two you want. Withdrawing consent removes billing data and ends paid access. Account deletion removes everything. - Read the warning panel in full before confirming. Neither action can be undone from this screen. ## Steps ### 1 — Open Settings, then Privacy ``` ┌─ Settings ───────────────┐ │ PREFERENCES │ │ Account │ │ Notifications │ │ ● Privacy ← here │ └──────────────────────────┘ ``` ### 2 — Check what is actually on file Three rows under "Data on file for billing". If they are all dashes, EOMA holds no billing identity for this brand and there is nothing here to withdraw. ### 3 — Set your marketing preferences Two tick boxes. Each one saves the moment you click it — there is no Save button. ### 4 — Withdraw consent only if you mean it **Withdraw consent…** turns into a confirm step. **Yes, withdraw and delete** performs it. ### 5 — For full account deletion, email us The link at the bottom of the page goes to the data deletion instructions. If it does not open, type the address `https://eoma.ai/data-deletion` directly. ## What you'll see ### The header — "Privacy & data" "Manage the personal data we hold for billing (KVKK / Turkish Personal Data Protection Law). You can withdraw your consent at any time." ### "Data on file for billing" Three rows: **Identity number**, **Phone**, **Consent recorded at**. The first two are masked — you see the first two and last two characters and dots in between. That is deliberate: enough for you to recognise it as yours, never enough for anyone reading over your shoulder. With nothing on file you instead get "No billing-identity data is on file for this brand. You can submit it when upgrading to a paid plan." ### "Commercial Message Preferences" "Manage whether you receive campaign, announcement, and promotional messages from EOMA. You can withdraw your consent at any time (ETK / Law No. 6563)." Two tick boxes — **Commercial messages by email** and **Commercial messages by SMS** — and a link to the ETK Notice. ### "Withdraw consent and delete data" An amber panel. It states plainly what happens: your identity number, phone and consent record are permanently deleted; an active paid subscription is cancelled, you keep access to the end of the current billing period, and no renewal is charged; you would need to submit identity again to subscribe in future. ### The line at the very bottom "Need to delete your entire account (not just billing data)? See data deletion." ### What EOMA holds, beyond this screen The billing rows above are the only personal data this screen can delete. Across the product, an EOMA account also holds your account and in-app settings, your name, email, phone and company details, any demo request and scheduling records, and the content and analysis output generated under your account. All of it is used to run the product you bought: measuring your brand, producing your recommendations, and billing you. Full detail is in the Privacy Policy at `https://eoma.ai/privacy`. ## Reading the numbers **"12••••••34" on Identity number.** Your national identity number as given to the payment provider, shown with the middle removed. Turkish payment rules require it for a card subscription, which is why it exists at all. It is masked on every load — the full value is never sent back to your browser after you first typed it, so no browser extension, screen recording or screenshot on this page can capture it. **"+90•••••••12" on Phone.** The same treatment: country code and last two digits only. **"Consent recorded at · 3 July 2026, 14:22."** The moment you agreed to EOMA holding the two values above. It is the legal anchor for the whole record, which is why withdrawing consent deletes the timestamp along with the data — the record is the consent, not just the fields. **"You keep paid access until 12 September 2026."** If withdrawing consent cancels a subscription, this is the date paid access ends. You are not refunded and you are not cut off early: you paid for this period and you keep it, and no renewal is taken afterwards. **"30 days."** The window inside which a full account-deletion request is processed once it reaches us, ending with a confirmation email to the address that sent it. Records we are legally required to keep — invoices and accounting documents — stay for the period the law requires, and aggregated data already shared with analytics providers is subject to those providers' own deletion tools. ## When it goes wrong | You see | Why | What to do | |---|---|---| | All three billing rows read "Not on file" | No billing-identity data is on file for this brand. Today's card form collects none of the three, so most brands look like this. | Nothing to withdraw. | | "Could not save your preference. Please try again." | The marketing tick box did not save; it flips back on its own. | Click it again. If it keeps reverting, contact support — do not assume it saved. | | "Couldn't revoke consent. Please try again." | The withdrawal did not complete. Nothing was deleted. | Retry. If it repeats, email us rather than pressing again. | | The "data deletion" link at the bottom does not open a page | The link currently points at an address that does not resolve. | Type `https://eoma.ai/data-deletion` into the address bar instead. The instructions are there. | | You withdrew consent and now cannot subscribe | Withdrawing deleted the identity record a card subscription needs. | Submit it again during a new upgrade. Nothing is permanently blocked. | | You want everything deleted, not just billing data | This screen only covers billing identity. | Email `info@eoma.ai` with the subject "Data Deletion Request", your registered name, the email you signed up with, and what you want removed. Expect a reply within 30 days, and expect to verify who you are. | ## Related - [Plan and usage](/docs/settings/plan-and-usage) — what cancelling a subscription does on its own - [Notifications](/docs/settings/notifications) — the action emails, which are separate from marketing messages - [Account](/docs/settings/account) — your sign-in email and two-factor sign-in --- # How your visibility is scored > Your AI Visibility score is eight separate measurements, blended. Here is what each one asks, and why a citation you earned is worth more than one you were handed. Source: https://eoma.ai/docs/concepts/how-visibility-is-scored Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Your score is eight separate measurements, each asking a different kind of question, blended into one headline. - Two of those measurements are about being *known*; five are about being *suggested*. They are weighted 30/70 in favour of being suggested. - A mention only counts as evidence if the question did not already say your name. Most brands' raw numbers are flattered by their own echo. ## The short version A single "visibility percentage" is a tempting number and a misleading one. "Do you know Yaseminkahve?" and "What is the best Turkish coffee brand?" are not the same question, they do not have the same kind of answer, and a system that scores both by looking for your name will tell you the first one is a win. So EOMA scores eight kinds of question separately, using the instrument each one deserves. **Three of them name you.** Recognition ("what do you know about Brand X?"), market position ("how does Brand X compare to its competitors?"), and sentiment ("what do people say about Brand X?"). These measure whether the engine *knows* you, and they are scored on an absolute basis — the engine either has real information about you or it does not. **Five of them do not name you.** Share of voice, share of voice in a niche, product recommendation, buying guide, and category discovery. These are the questions a buyer actually asks before they have heard of you, and they measure whether the engine *suggests* you. They are scored against the best-performing brand in your market rather than against a perfect 100. The headline blends the two: **30% knows-you, 70% suggests-you.** The weighting is deliberate. Being known is table stakes; being recommended to a stranger is the thing you are buying. EOMA never collapses the pair into one figure on screen, because the gap between them is usually the whole story. A brand that scores 84 on knows-you and 0 on suggests-you is famous and absent — and that diagnosis is invisible if you only read the blended number. ## How it works 1. **Each question is answered by the instrument that fits it.** A question that says your name cannot be scored on rank, because rank is meaningless when the answer was handed the subject. A question asking for a shortlist is scored on *share of the list*, not raw position — being third of three is the bottom of the shelf, third of twenty is the top. A comparison where the engine named no competitor at all is excluded rather than counted as a win. 2. **"I have no information about that brand" scores zero.** The engine disclaiming you is not a mention. This sounds obvious and it is the single most common way visibility numbers get inflated across this whole category of tool. 3. **Being absent scores zero. Not being measured yet scores nothing at all.** Zero means "we asked, and you were not there". Blank means "we have no evidence". Sentiment in particular reads *not measured* rather than a neutral score for most brands, because reading tone from an answer is a separate, paid analysis that has only been run for some brands. A neutral face would claim people feel nothing about you, which nobody measured. 4. **The five suggests-you axes are scored against your market's real ceiling.** Nobody reaches 100. The best-performing brands EOMA has measured top out around 47% on share of voice; most markets peak in the low teens. Scoring against an absolute 100 would tell the category leader it was failing. So each of those axes is expressed as a fraction of the best rival's score on the same instrument, measured on the same set of questions. 5. **The score covers 28 days, not one run.** It pools every answer from the last four weeks — typically several runs and well over a hundred answers. If a brand has had no runs in 28 days, the score reads **not measured**. The window never stretches to reach older data, so you never see a stale number wearing a current date. ## What this means for you **Your citation count is probably flattered by your own echo, and the fix changes the number a lot.** When a question says "tell me about villa sepeti", any answer citing villasepeti.com is the engine quoting you back to yourself. That is not evidence anyone found you. EOMA separates the two, and ranks the sources board on *earned* citations only — the ones from questions that never mentioned you. Measured across every brand with a website: villa sepeti showed 76 citations, 71 of them handed over, **5 earned**. Yaseminkahve: 31 shown, **1 earned**. Six of the ten brands measured had **zero** earned citations. If you are reading a raw citation count anywhere else, assume most of it is echo until you check. **A brand's own row survives at zero earned, on purpose.** "Cited 30 times, never unprompted" is a finding worth acting on. Dropping the row would render it as the flatly different "not cited at all". **Low suggests-you with high knows-you is a content problem, not a brand problem.** The engines know who you are; they are not reaching for you when somebody asks the buying question. That is the gap your weekly actions are aimed at. **The score moves slowly and that is correct.** Publishing something today does not change what an engine says tomorrow. Judge the direction over weeks. ## Common misunderstandings | People assume | Actually | |---|---| | "One percentage describes my AI visibility." | It cannot. Eight categories are measured on four different instruments, and one figure that averages them hides which half you are failing. The knows-you / suggests-you pair is the smallest honest summary. | | "100 is the target." | It is unreachable on the five suggests-you axes and always has been. The measured market leaders sit between 10% and 47%. Your number is a fraction of what the best brand in *your* market achieves. | | "A citation is a citation." | Only if the question did not hand it to you. An answer to "tell me about my brand" citing your own site proves nothing about discovery. | | "Sentiment is blank because nobody has opinions about us." | Blank means the tone analysis has not been run for your brand. It is a separate paid step, and EOMA leaves it empty rather than substituting a mention rate wearing the word "sentiment". | | "My score dropped, so something got worse." | Check whether the window moved. The score pools 28 days; a run falling out of the window changes what is being averaged. Read several weeks, not two points. | ## Related - [What a run is](/docs/concepts/what-a-run-is) — where the answers being scored come from - [Competitors and share of voice](/docs/concepts/competitors-and-share-of-voice) — who sets the ceiling your score is measured against - [AI provider visibility](/docs/pages/ai-provider-visibility) — the screen where this score lives --- # What a run is > EOMA puts your questions to four AI engines on a schedule. Here is why the same question is asked over and over, and why one run never proves anything. Source: https://eoma.ai/docs/concepts/what-a-run-is Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - A run is one round of your questions being put to ChatGPT, Claude, Gemini and Perplexity, and the answers being recorded. - Paid plans run once a week — four times a month. Not every question goes out in every run. - These engines do not answer the same question the same way twice, which is exactly why the measurement is repeated rather than taken once. ## The short version An AI engine is not a database. Ask ChatGPT "what are the best villa rental sites in Antalya?" three times and you can get three different lists. Nothing broke; that is how the technology works. It means a single answer tells you almost nothing, and a single answer treated as a measurement produces a number that swings for no reason and gets acted on anyway. EOMA's answer to that is repetition. Your questions go out on a schedule, to four engines, week after week. Each individual answer is an anecdote. Several hundred of them, pooled over 28 days, is a measurement — and the same pool measured again next month is a trend. That is the whole reason the product accumulates rather than reporting. In your first fortnight you have a snapshot. By week eight you have a direction, and the direction is the part that tells you whether the work you did was worth doing. ## How it works 1. **A schedule fires.** On every paid plan, the AI visibility collection runs **once a week** — four runs a month, on the same weekday each week. The spacing is deliberate: your monthly question allowance is spread across those four runs instead of being spent in one enormous tick, so a number always reflects several points in time rather than one afternoon. 2. **A quarter of your question set goes out, not all of it.** Each run asks your monthly allowance divided by four — so a Starter run puts out about 8 questions, Growth 25, Scale 50 — and the remainder waits for the next run. Four runs, one month, your whole allowance asked exactly once. If you track fewer questions than your plan allows, the run is sized on what you actually track, never on the plan's ceiling. 3. **Each question is put to four engines.** ChatGPT, Claude, Gemini and Perplexity, with the same wording, adjusted only where an engine will not accept a setting — Gemini takes no location fields, so the location has to sit in the question itself; Perplexity accepts a country but not a city. Each question is also asked in two modes, one where the engine may search the web and one where it answers from what it already knows. That distinction is frequently the finding: a brand the engines can *look up* but do not *know* is a different problem from one they cannot find at all. 4. **No question is asked twice until every question has been asked once.** The rotation picks the questions that have gone longest without being asked, so a question you added last week is asked before one asked yesterday. That is a guarantee about coverage rather than a waiting period: it is what makes "four runs cover your whole set" true rather than approximately true, and it is why the rotation cannot get stuck on your five favourite questions. 5. **Answers are added, never replaced.** A question asked in July and again in August produces two records under the same question, not one overwritten record. That is what makes a per-question history possible, and it is why editing a question's wording does not detach it from its own past. 6. **The scoring runs afterwards.** Answers are read for who was named, which sources were cited, and whether the engine actually knew anything. Competitor names are folded together first — "Airbnb" and "AirBnB" are one rival — and only then is anything scored. Scoring before that step would count one competitor as two and quietly deflate everything. ## What this means for you **Not every active question is asked every month, and that is not a fault.** Your active set is a rotation pool, not a promise. If you add thirty questions mid-month, some of them will not be reached before the month turns over. Some may wait several months. There is no error, no warning and no health check for this, deliberately — a warning that fires constantly and correctly is one everybody learns to ignore. **Turning a question off and on again does not push it to the front of the queue.** The 30-day clock survives deactivation. If you switched a question off five days after it ran, switching it back on does not buy it an extra send. **A small active set gets re-asked, not under-spent.** If you narrow to five questions, those five are asked repeatedly. Your cost does not change; your data just gets repetitive. **One run is never evidence, and neither is one day's change.** If your number moved and you changed nothing, the most likely explanation is that the pool moved: a run entered the 28-day window and an older one fell out. Compare weeks. **If nothing has run in 28 days, EOMA says so rather than showing you an old number.** A blank score with "not measured" is the honest state. Data frozen weeks behind its schedule is worth reporting to us. ## Common misunderstandings | People assume | Actually | |---|---| | "Every question I have is asked every time." | Each run takes a slice sized by your plan. The whole set is covered across a month of runs, and a question added late may wait. | | "The answer changed, so the tool is unreliable." | The engines genuinely answer differently between asks. That variance is the reason for repeated measurement, not evidence against it. | | "A run is a report being generated." | A run is EOMA spending real money asking real engines real questions. It produces new evidence, which is why it is paced rather than run on demand. | | "I published a fix, so the next run should show it." | The engines re-read the web on their own schedule. Weeks is the normal delay between a change going live and an answer changing. | | "More runs would give me a better number." | Beyond the current cadence, more runs mostly buy noise. The 28-day pool already holds several runs and well over a hundred answers for an active brand. | ## Related - [Why your prompts matter](/docs/concepts/why-prompts-matter) — what actually goes out in a run - [How your visibility is scored](/docs/concepts/how-visibility-is-scored) — what happens to the answers afterwards - [Plans and limits](/docs/start-here/plans-and-limits) — how many questions your plan sends --- # Why your prompts matter > Your prompts are the questions EOMA puts to the AI engines. What you write in your brand profile becomes those questions — and the wrong question measures the wrong thing. Source: https://eoma.ai/docs/concepts/why-prompts-matter Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - A prompt is one whole question a buyer might ask an AI assistant — not a keyword. - What you write in your brand profile is what the questions are built from, so a vague profile produces questions that measure somebody else's market. - You can add your own questions, in any language. EOMA classifies and locates them for you rather than making you learn the categories. ## The short version Everything on your AI Visibility screen is downstream of one thing: the list of questions being asked. Change the questions and every number changes, because you are measuring a different market. This is the part of EOMA most worth your attention, and the part founders usually skip. A question like *"best coffee"* measures a global commodity market you will never appear in. *"Where can I buy Turkish filter coffee online in Germany?"* measures the market you actually sell into. Both are valid questions. Only one of them tells you anything. EOMA writes your first set for you, from your website, your category and your markets. Your plan sets two numbers, and they are different on purpose: how many questions you can **keep** (50 on Starter, 200 on Growth, 300 on Scale, unlimited on Enterprise) and how many get **asked each month** (30 · 100 · 200 · 400). Keep more than you ask and the extra questions are not wasted — the rotation reaches them on a later cycle, oldest first. Then you can edit, deactivate, and add your own. A prompt is a **question**, not a keyword. It is asked as a person would type it, and the engine answers in prose. That is why the questions run up to 500 characters and why they are full sentences. ## How it works 1. **Your brand profile becomes the source material.** Your website, the competitors you named, your markets, your industry and your footprint are all read when the question set is written. This is why the setup questions during onboarding are not paperwork — a brand that leaves "SaaS & Software" sitting in the industry field because it was pre-filled gets a question set about SaaS. Edit it in [Brand profile](/docs/settings/brand-profile). 2. **Every question gets a category, and the category picks the instrument.** There are eight: recognition, market position, sentiment, share of voice, share of voice in a niche, product recommendation, buying guide, and category discovery. The category decides how the answer is scored and whether the question counts towards your share of voice. A miscategorised question is a wrong number arrived at silently, which is why nothing is ever defaulted. 3. **Questions you add are classified, not guessed at.** When you submit your own questions, EOMA reads them against the same specification its generator writes against, and assigns the category the generator would have used. It then checks that answer mechanically: three categories *require* your brand name to appear in the question, five *forbid* it, and sentiment allows either. A classification that breaks its own rule is sent back for a second attempt, and a question that still cannot be placed is rejected with a note rather than stored wearing a made-up category. 4. **Geography is looked up, never inferred from your words.** The country a question is measured in comes from your brand's own market settings. The city field, where it is used, describes **where the person asking is sitting** — not what they are asking about. Somebody asking about Antalya villas is usually in Istanbul. An earlier version of this read the city out of the question text and tagged that Antalya question as an Antalya searcher, which measured a narrower question than the customer had. 5. **Any language works.** Turkish, German, Arabic — the question is asked as written and never translated. Language changes neither the category nor the market it is measured in. A Turkish question is classified exactly as its English equivalent would be. 6. **Adding questions tops up; it never replaces.** Regenerating your set inserts only questions that are genuinely new. Existing questions, their on/off state and their whole answer history are untouched. Editing a question's wording keeps its identity, so its history continues on the same line with an "edited here" marker rather than starting over. ## What this means for you **Your plan limits how many questions can be *active at once*, not how many you may own.** Starter tracks 100 active, Growth 200, Enterprise 400. If you are at the cap, activating another question asks you to switch one off first. Questions you switch off keep every answer they have already earned. **Your own questions are capped at 100 per brand, on every plan**, and at 25 per upload. That pool is the same on every tier because it is a size limit, not a feature you are being sold. **A question that names your brand and a question that does not are measuring different things.** "What do you know about my brand?" tells you whether the engine has correct information. "What are the best X?" tells you whether it recommends you. Both belong in your set. Only the second one moves the number buyers care about. **If your brand has no market set, your questions cannot be stored.** The country comes from your market settings, so a brand with none gets a plain refusal rather than a silent default. Set your market in Brand profile first. **Watch what your first set actually asks.** It is written from your website, and a website that undersells one of your lines will produce a question set that never measures it. Half an hour here changes every number on every other screen. ## Common misunderstandings | People assume | Actually | |---|---| | "Prompts are keywords." | They are whole questions in natural language, up to 500 characters, answered in prose. Keyword thinking produces questions nobody would ever type into an assistant. | | "I should write my questions in English so the engines understand." | Write them the way your buyers ask them. The engines answer in the language of the question, and that answer is what your buyers would see. | | "I added a question, so it will be asked in the next run." | Active means eligible, not scheduled. Each run takes a slice sized by your plan, and a new question waits its turn in the rotation. | | "I should include my city in the question to measure my city." | The location is a setting, not text. Where the *asker* sits is a separate thing from what they are asking about, and confusing the two narrows what you measure. | | "Deleting a question deletes its data." | Switching one off keeps every answer it earned. Its history stays readable and stays in the pool the score is computed from until the window moves past it. | ## Related - [Brand profile](/docs/settings/brand-profile) — the facts your question set is written from - [What a run is](/docs/concepts/what-a-run-is) — how your questions actually go out - [How your visibility is scored](/docs/concepts/how-visibility-is-scored) — what the eight categories do to the numbers --- # Competitors and share of voice > Share of voice is your slice of every brand the AI engines named. EOMA shows two competitor sets — the ones the engines chose, and the ones you did. Source: https://eoma.ai/docs/concepts/competitors-and-share-of-voice Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-08-14 --- ## In 10 seconds - Share of voice is your slice of all the brand mentions across your answers — yours against everyone else's. - EOMA shows two different competitor sets: the brands the engines named on their own, and the list you keep. - Your score's ceiling is set by the strongest rival in your market, not by a perfect 100. ## The short version When someone asks an assistant "what are the best options for X?", the answer names two or three brands. Share of voice counts how often one of them was you. The uncomfortable part is who else is in the ring. The engines cite whoever answered the question well: a directory, a marketplace, a forum thread, a newspaper, a rival three times your size. For the purpose of that one question, those are your competitors, whatever your competitive analysis says. Villa sepeti's ring holds around seventy names; the nineteen it tracks deliberately are a small part of the picture. So EOMA keeps both views and lets you switch between them. **What the AI named** is the open world — every brand the engines mentioned, whether or not you consider it a competitor. This is the default, and it is the honest answer to "who is winning this conversation". **My tracked list** is the set you chose. It is smaller, it is stable, and it is the right view when you want to answer "am I beating the three companies my board asks about". They will disagree, and the disagreement is not a bug. One customer wrote in unable to reconcile "#3 of 12" on one card with a ring of seventy rivals on another; both numbers were right, and they were answers to different questions. ## How it works 1. **Every answer is read for the brands it named.** Names are folded together first — "Airbnb" and "AirBnB" are one company, not two — so a rival that spells itself three ways does not appear as three weak competitors instead of one strong one. 2. **Your slice is computed from mention instances, not from answers.** If one answer names a rival twice, that is two instances. This is worth knowing because it makes the donut's percentage and the headline mention rate two different figures describing the same reality on different denominators. 3. **In the open-world view, the chart makes room for the tail.** The top five rivals always get a slice, so one dominant leader cannot erase rivals two through five. Beyond those five, a rival needs at least 2.5% to appear, and the chart shows at most ten. Your own slice is always lifted with a ring and stays lifted while you hover elsewhere — you never lose your own anchor. 4. **In the tracked view, none of that filtering applies.** No cap, no threshold. A list you chose is already bounded, and a tracked rival with a single mention is the finding, not noise worth hiding. 5. **Rivals the engines never named are listed on the back of the card.** A zero has no arc — on a donut, a competitor with no mentions looks identical to one that was never tracked. Flip the card and they are named explicitly. 6. **The strongest rival sets your ceiling.** The five demand axes of your visibility score are each expressed as a fraction of the best rival's score on the same instrument, measured on the same questions. That is why the back of this card and the donut on the front belong together: one shows your slice, the other shows the number that slice is measured against. ## What this means for you **Expect names you did not plan for.** Directories, marketplaces and publishers routinely out-cite operating businesses, because they answered the question in a form the engine could quote. Treating them as "not real competitors" does not recover the answer. **A tracked rival at zero mentions is a signal about them, not about your data.** It usually means the engines do not know them either — occasionally useful, and worth checking before you assume the tracking is broken. **Your percentage moves when you filter, and that is arithmetic, not measurement.** Percentages are computed after any category filter, so switching a filter changes your own share. Compare like with like. **Edit your tracked list from the chart, not from Settings.** The "Adjust list" control sits at the top right of the donut area and is always visible, including when you have no competitors tracked at all. **A single strong rival can be the whole ceiling.** In thin markets, one company above the observation floor sets the number your demand axes are graded against. Where no rival clears that floor, EOMA says there is no ceiling rather than inventing one. ## Common misunderstandings | People assume | Actually | |---|---| | "The competitor list is wrong — half of those are not our competitors." | In the open-world view it is not a list of your competitors; it is a list of who the engines named when your buyers' questions were asked. Switch to the tracked view for the list you chose. | | "Two screens show different competitor counts, so one is broken." | They answer different questions. The tracked card counts your list; the donut counts everyone named. Both numbers can be right at once. | | "A rival with no slice is not being tracked." | On a donut, zero is invisible. Flip the card — rivals the engines never named are listed there by name. | | "Getting to 100% share of voice is the goal." | No brand in any market EOMA has measured is close. The measured leaders sit between roughly 10% and 47%. Your target is the ceiling in your market, and that number is on the card. | | "Adding competitors will improve my score." | Tracking a rival changes what you can see, not what the engines say. The score's ceiling comes from the strongest rival the engines actually named, tracked or not. | ## Related - [How your visibility is scored](/docs/concepts/how-visibility-is-scored) — how the ceiling turns into your number - [Share of Voice](/docs/pages/share-of-voice) — the screen where this lives - [Competitors](/docs/pages/competitors) — the search-side competitor view --- # The life of an action > An action starts as a recommendation, gets started, worked, drafted, approved and published. Here is every state it can be in and whose turn it is. Source: https://eoma.ai/docs/concepts/life-of-an-action Product: EOMA — AI search visibility for brands (https://eoma.ai) Last updated: 2026-10-07 --- ## In 10 seconds - An action moves through a small, fixed set of states, and every one of them is either waiting on EOMA or waiting on you. - Pressing **Start** hands the job to EOMA and cannot be undone. It uses one of the jobs your plan has EOMA do for you. - The last step — approving where the work goes — is always yours, because EOMA will not publish to your live site without being told to. ## The short version Your board is a short list of specific jobs: this page needs writing, this claim needs a source, this profile needs correcting. On Starter it holds six fixed tasks, each coming back on its own schedule; on Growth and above, EOMA also turns every week's findings into new ones. The board is cut by **whose turn it is**. *Waiting for you to start* holds the offers nobody has taken. *EOMA does these* holds the work EOMA has claimed and is working on. *Waiting for your review* holds everything that has stopped until you decide something. EOMA can do any action on the board, and you can do any of them yourself with the steps inside it. Nothing happens on a row until you press **Start**. Before that it is an offer, not work in flight. This distinction was worth rebuilding the board over: rows sitting under a header saying "EOMA is handling these" while nothing had been charged and nothing was running read as progress that did not exist. Once started, the row is claimed, an agent works it, and it comes back to you with something to approve. The whole rest of this page is the vocabulary for the middle part. ## How it works 1. **The row is generated.** A fixed task arrives on its own schedule and comes back after you finish it, never more than one returning in the same week. On Growth and above, EOMA also writes new recommendations every week, up to ten open at a time, and tops the list up as you close things, so the board never becomes a backlog nobody reads. 2. **You press Start, and EOMA takes the job.** A confirmation says what happens and what it uses: EOMA starts working and comes back with a draft in about two days, nothing is published until you approve it, and the job is counted on your plan. On Free and Starter it says how many jobs will be left; on Growth and Scale it says the job holds one of the places running at once until it finishes. There is no un-start. If you would rather handle a row yourself, open it and follow its steps, or use "I've already done this"; neither uses anything. 3. **An agent works it.** It reads your brand data, decides whether the recommendation is still valid, prepares the work, and posts progress onto the row as it goes. If it needs a fact only you have, it stops and asks. 4. **A draft comes back for approval.** For most kinds of work the agent produces something concrete — a page, a schema block, a correction — and parks it. Nothing goes anywhere until you approve. 5. **You choose the destination.** Approving opens a fork, because the two destinations do not promise the same thing: | Destination | What happens | |---|---| | Your site (WordPress) | The draft is written to your live site as it stands. Visitors see it immediately, and the action closes itself as done. | | Your code repository (GitHub) | A pull request is opened. Nothing is written directly, you see the change line by line, and nothing goes live until you merge. | 6. **The row closes.** On the live-site path it closes automatically. On the pull request path, someone has to confirm the change is actually live. **The states a row can be in**, and what each one means in plain terms: | The board says | It means | Whose turn | |---|---|---| | Waiting for you to start | An offer nobody has taken. Nothing has been used. | Yours, if you want it | | Working on it | An agent is actively progressing it | EOMA | | Needs your input | The agent asked a question it cannot answer for you | Yours | | Review the draft | Work is finished and parked, waiting on your approval | Yours | | Review & merge | A pull request is open on your repository | Yours | | Connect to publish | The work exists but there is no connected site to send it to | Yours | | Approved, not yet out | You approved and the publish has not fired yet | EOMA | | Published, not closed | It went out and nobody has confirmed it | EOMA | | Done / Failed | Terminal. Stays in your notifications for seven days, then it is history | Nobody | An unstarted row never waits on anybody. If a row is asking you for something, it has been started. ## What this means for you **Start is irreversible and it takes a seat.** There is no un-start. On Growth three jobs can be in flight at once (six on Scale) and a seat frees the moment one reaches done or is removed. On Free EOMA does 2 jobs for you and on Starter 3; that number is for the life of the plan, and a job does not come back when it finishes. When Starter's are used, Start says so and points you to the action's own steps and to the plans; on Free the other cards wait behind a lock until you upgrade. The seat meter counts the rows in the in-progress band, so the two cannot disagree — with one short-lived exception: work already in flight before 2026-09-08 does not take a seat, so until those jobs finish the band can hold more rows than you have seats. **"Connect to publish" ranks below anything already finished.** A parked draft on a brand with no connected site still says *review the draft*, not *connect something first* — because you can open it, export it as Markdown, HTML or PDF, copy it as a prompt, apply it yourself, and mark the row done. Connecting a site makes that automatic; it is not a gate on getting the work. **When you merge a pull request, EOMA does not notice on its own yet.** There is no webhook watching your repository, so the row stays In progress until somebody closes it. One customer's pull request sat open for eight days looking like work in flight. If you merge something, tell us or close the row — this is a known gap and it is being fixed. **A row that says nothing for days is worth asking about.** The board shows "taken on N days ago · no update yet" for exactly this reason. **Removing a suggestion is allowed, and only before it starts.** A row you have started cannot be dismissed; a row you have not can, and it can be restored. ## Common misunderstandings | People assume | Actually | |---|---| | "The recommendations are EOMA's to-do list, so they will get done." | Nothing is claimed until you press Start. Before that a row is an offer, and it waits under **Waiting for you to start** until you take it or do it yourself. | | "Approving publishes it." | Approving asks where it should go. The live-site path publishes immediately; the repository path opens a pull request and stops. | | "If EOMA cannot publish it, I cannot use it." | Every draft waiting for you can be taken away as Markdown, HTML or PDF, or copied as a prompt, from its last step. Before a draft exists, the action's instructions can be copied as a prompt for your own AI. | | "In progress · 3/6 means I have used three of my six." | It means three are running and your bank holds six. The note beside it says how many remain. | | "I can start six things and pause four." | Starting uses the job at the moment of the click. There is no pause and no refund. | | "Failed means EOMA broke." | Often it means the work could not be completed as scoped — a site that blocks access, a fact nobody could confirm. The row carries the reason. | ## Related - [Actions](/docs/pages/actions) — the board itself - [Actions](/docs/pages/actions) — where the week's list is worked - [Plans and limits](/docs/start-here/plans-and-limits) — how many executions your plan includes