One in eight businesses AI names is ever its first pick

Short answer: when an AI assistant answers "the best {trade} in {city}", being named is almost the easy part - it lists seven or eight businesses. Being named FIRST is the scarce slot. Across France, the UK and Brazil, the web-searching assistant named 4,836 different businesses; only 643, about one in eight, were ever its first pick.

You run the check and there you are: the assistant mentioned your business. Relief - you are in it. But a buyer reading that answer does not read all eight names. The eye stops near the top. So the real question is not whether you are named; it is where. First, or sixth?

Being named is nearly the floor. The first line of the list is the part that is fought over.

What was measured

Waseit's studies ask, in each city and trade, the question a customer types: "what are the best {trade} in {city}?". We read the businesses each answer names, in the order it gives them, using the explicit rank the answer itself carries. Then, for each country, we counted how many different businesses were named at all, and how many of them ever reached first position. The published figure is anchored on Perplexity, the one of the three assistants that searches the web before answering: its lists are genuine businesses (our business-vs-advice filter rejects just 0.9% of them). Perplexity answered only in France, the UK and Brazil, so the figure covers those three markets: 682 answers to one phrasing of the question.

We anchor on Perplexity for a reason. Ask ChatGPT or Gemini "the best plumber in Lyon" and they often answer with a procedure - check reviews, ask around, verify qualifications - which our extraction counts as names. We cannot cleanly separate that advice from real businesses across sixteen thousand entries, so we publish no precise figure for them. When ChatGPT answers with advice instead of a name is a measurement of its own.

Why does being named count for so little?

Because the list is long and the top is short. In a typical answer the assistant names close to eight businesses - so a mention is nearly the floor. One of them is first. Across the three markets, 4,836 businesses cleared the first bar (being named at all); 4,193 of them were never anyone's first pick. The share that reached the top is strikingly flat from market to market - 14% in France, 13% in the UK, 13% in Brazil - which says this is structural, not a quirk of one country.

And the same shape appears for the other two assistants, around one in seven, though their lists mix in advice we cannot fully strip - so we read it as an echo, not a precise number.

What it means if you are the business

Being mentioned by an AI assistant is no longer the finish line - it is the starting line. For 97% of local businesses the problem is still not being named at all; but once you clear that bar, a second one waits, and it is narrower: being the name a buyer actually reads. Treat visibility as a rank, not a yes or no. If the assistant names you eighth, you are visible and unchosen. The useful question becomes: on each assistant, where does it put me - and, for the one that searches the web, do the pages it reads say I belong near the top?

And your business - where do the assistants put you?

That is what Waseit's free audit shows: it asks ChatGPT and Gemini the eight questions your customers actually type, in your language and for your market, and gives you back, in about 60 seconds, a score out of 100, the competitors named in your place, your position assistant by assistant - not just whether you are mentioned, but where - and which assistant actually searched the web rather than answering from memory. Free, no account, no credit card; an email address only unlocks the competitor names and the action plan. It is a dated snapshot, not a ranking, and we promise no position.

What this does not prove

• The published figure is Perplexity's, on France, the UK and Brazil - the only markets it answered. ChatGPT and Gemini show the same roughly one-in-seven shape, but their answers mix in how-to advice our extraction counts as names, so we give them as an order of magnitude, not a precise figure.
• Reading names out of an answer is imperfect even for Perplexity: we keep a capped number of names per answer, and a "first" slot can occasionally be a directory (Triider, Which? Trusted Traders) rather than a single business. We counted positions, not the quality of what fills them.
• "First" here is the top rank the assistant itself gave within one answer. It is not a market ranking, and nothing says the first name is the best business or that the others are worse.
• One capture, between 31 July and 16 October 2026, through the assistants' programming interfaces, which approximate the consumer apps without reproducing them. Answers change from one day to the next; this is a photograph of them.
• We measured the share of named businesses that ever reached first, not how often any single business holds first. A name can top one city's answer and sit mid-list in the next.