Field Notes · 5 min read ·

1,000+ brands, 40,000+ answers

For more than a thousand brands we simulated the questions their customers ask: do they get recommended, is the name recognised, does AI speak well of them, where do they land against competitors. Here is what 40,000 answers said.

Türkçe oku

For more than a thousand brands, we simulated the questions their own customers ask. Across dozens of different industries we put buying-intent questions to the four most-used AI engines and went through more than forty thousand answers one at a time, looking at whether the brand gets recommended, whether its name is recognised, whether the AI speaks well or badly of it, and where it lands next to its competitors.

The questions look like this. "Which clinic or doctor is best for a hair transplant in Ankara?" "Which dealer in Istanbul is most trusted for imported cars?" "Is it safe to get medical treatment in Turkey, and which providers would you recommend?"

The first thing forty thousand answers told us was this: three out of four brands do not appear in any of them. Not once.

The engines are not stuck when the brand is missing, either. They list an average of seven competitors by name and produce a confident, complete answer in which the only thing missing is the brand itself.

A healthy website does not save you

The first thought a brand has on hearing this is usually the same one: maybe AI crawlers cannot read my site.

It is a fair suspicion, and the data points elsewhere. Of the brands whose site does not load at all, 88% are invisible; of the brands whose site is technically perfect, 73% are invisible too.

Fixing crawlability moves you from 88 to 73, which is a real gain that leaves the other 73 points completely unexplained. The problem is also rarer than the conversation suggests, since only about one site in twenty turns out to have a structure a bot cannot read.

None of that makes crawlability irrelevant. The internet is being built a second time for machines and that side of it is still raw. But it is not why a brand goes unrecommended.

The same three mistakes, in every industry

The industry changes, the city changes, the budget of the person asking changes. The three problems do not.

Horizontal bars: no independent record outside 84%, page does not say who it is 77%, nothing quotable written 70%, all three at once 45%. The three problems brands share No independent record outside 84 Page does not say who it is 77 Nothing quotable written 70 All three at once 45 % From the report findings of more than a thousand brands.
A dental clinic, a furniture shop, a hotel, a medical manufacturer. The order barely moves.

Nobody outside is talking about you

In 84% of brands the picture is identical: not one independent source mentions the brand. No industry directory, no listing, no press, no reviews.

This is the hardest part to accept. An AI answering a recommendation question behaves like a friend answering in a group chat: it repeats the names other people vouched for, not the names that advertised. Whatever you write on your own site, you are still the one praising yourself.

If nobody speaks for you, the AI has nothing to say about you.

Your page does not tell a machine who you are

In 77% of brands the page title is either generic or missing, with no H1, no description and no structured data.

The version we see most often is a title where the brand name sits second, behind a category, along the lines of "Medical Products and Wholesale Medical Supplies." For a person that is fine, because they can see the logo above it. For a machine that line is the entire identity, so your page reads like a shop with the category painted on the awning and your name nowhere on the door.

You have not written anything quotable

In 70% there is no blog, no FAQ, no buying guide and no comparison page. Sites are perfectly clear about what they sell.

But an AI is not answering "what do you sell." It is answering "which one suits this person," and to do that it needs a sentence it can lift and quote, which a product page does not contain.

A product page sells. It does not answer.

The good news is that all three are doable

Nearly half of the brands carry all three at once, and the scores say the same thing: nine out of ten land in the bottom band. If you have not checked, the odds are that includes you.

None of this is bad news, though, because all three are content and registration work rather than infrastructure work, and not one of them requires rebuilding a website.

The order is clear too: first make the page introduce itself, which is an afternoon; then write something quotable, which takes a few weeks; and last, get registered outside, because that is the part you cannot do alone and it takes months. Most brands start at the far end and put the slowest job first.

Then there is the next question: the sale does not close at the first prompt.