AI visibility for salons: what being named means and how to measure it

What does it mean for a salon to be named?

It means your salon's name appears in an AI assistant's answer to a client's question, and there are degrees. Mentioned: the name is there. Recommended: the assistant puts the salon forward as a choice. First: yours is the name the client reads before any other. Then there is the company you keep, the other salons in the same answer, who are the ones the client compares you with. Our method records three of them for every answer: whether the salon was named, whether it came first, and who else was named. Its visibility score is the share of the twelve questions in which the salon was named, from 0 to 100, with first place and the other salons reported beside it rather than weighted in.

A salon needs two more things written down. What you were named for: a salon can come back for "curly cut" and be missing for "balayage", or come back with one stylist's name attached. And whether the facts were right. An answer that names you with last year's prices, a closing time you changed, or a stylist who left is a problem you can see only by reading it, and the fix is on your site, not in the assistant.

The unit is one question put to one assistant. There is no single ranking of salons to look up: each assistant answers each question in its own way, and a salon can be in Claude's answer and absent from Gemini's on the same afternoon. That is why the measurement is a fixed set of questions and not a search you run once.

The questions a salon's panel should hold

A panel is a fixed list of client questions in the words clients use, asked the same way each time. For paying clients ours holds twelve, set once for the trade and the location and then left alone so that one reading compares with the next. For a salon, the questions come from the front desk: what people ask on the phone, in messages and at the counter before they book.

Write most of the panel as the first five kinds, which come from clients who have not heard of you, and one or two as the last. Put the place in the words of the question ("in kits", "near leslieville") rather than leaving it to "near me", so every run asks about the same place wherever you happen to be when you ask. And use the names clients use, for the neighborhood and for the service. The near-me article and the article on specialty pages cover both from the site's side.

How to run it without fooling yourself

Every question goes to ChatGPT, Claude, Gemini and Perplexity three times each, because the same question can get a different answer on each run; the same questions go again the next month. Why a single run misleads, and how to tell a change from noise, is in how to measure AI visibility. Two cautions are particular to an owner asking about their own salon.

Ask as a stranger would. An assistant you have used for months to draft your captions and price lists may already connect you with your salon. Where it offers the choice, switch memory and personalization off, or ask from an account you do not use for work, so the answer is closer to the one a new client gets.

And fix the place. "Near me" means somewhere different depending on where you ask from, the salon on a Tuesday and your kitchen on a Sunday, which makes two readings hard to compare. Naming the neighborhood in the question takes that variable out. A panel whose questions all name a place also tells you which neighborhood names work for you and which do not.

What to write down for each answer

Every answer gets the standard record: named or not, recommended or not, first or not, who else was named and in what order, and the whole answer pasted with the date, the assistant and the run. The format, and why the verbatim copy matters, is in setting up prompt monitoring. Fill it in the same way each time, by the same person, so that a reading in March means what a reading in June means. A salon adds three fields of its own.

• What you were named for: the service, the stylist, the hair type, in the answer's own words.
• Every fact stated about you (price, hours, address, stylists) and whether it is right today.
• The pages the assistant cited, where it shows its sources, which tells you where the facts came from.

The facts field is where a salon finds its quickest fixes. If an answer names a stylist who left, look for the page that still lists them: an old team page, a directory entry, a booking profile nobody closed. Correct the page, and the next readings show whether the answer followed. The same goes for a price from an old menu or hours from before a change.

Reading a salon's first baseline

Expect the first reading, your baseline, to be thin. Alongside our audit of 9,483 small business websites, run across 40 trades and 90 metros in the US, Canada, Europe and the UAE from late August to early September 2026, we put one buyer question from each business's trade and metro to Claude, with web retrieval on, for 9,240 of them. 14.9% of them were named; among the 5,928 answers that gave a clearly differentiated list rather than a hedge, 17.7%.

That was one question, one assistant and one day across all trades, not a salon figure and not a rate. It does suggest that a first reading which rarely names you is the ordinary starting point rather than a verdict on the salon.

Read the baseline question by question. For each question where another salon was named and yours was not, open that salon's site and write down what it states that yours does not: a page for the service, a price, the stylists by name, hours for the evening the client asked about. That list is your competitive analysis, and it is written in facts rather than impressions, which is what makes it something you can act on.

Note who the other names are, too. Across the whole probe, 99.4% of the names the assistant gave were actual businesses and 0.6% were directories or aggregators. If your panel follows that pattern, the names beside yours will be other salons rather than booking marketplaces, and your share of voice is a share of a short list of salons.

Last, write down the state of your site on the day of the baseline: which pages existed, where the menu lived, whether the stylists were named, whether the hours were in text. A later reading means more when you can put it next to what changed in between.

What a reading can tell a salon, and what it cannot

It cannot count bookings. A client who heard your name from an assistant books through your app or calls, and nothing in the answer tells you so. Ask instead. Add "an AI assistant (ChatGPT, Claude, Gemini, Perplexity)" to the question on your booking form or new-client card about how they found you, and read that count next to the panel.

It covers the questions in the panel and the assistants they are put to, which in our readings is one, Claude, and it compares only with itself: a reading taken with different questions is a different instrument. What it does tell you is whether you are named, for what, beside whom and with which facts, and whether that moves after you change your pages. The pages are the part you control. Which to change first is in the fix order for a salon, and the guide for hair salons lists what each page has to state.

You can run the panel by hand with a spreadsheet. To have a reading taken for you, the free visibility check takes your website and an email address, puts the questions to one assistant, Claude, and a person sends back a written result within 48 hours. Monitoring puts a fixed panel of twelve questions to Claude every month for $290 a month, which you can cancel in any month. Neither changes what an assistant says, and we do not promise a place in any answer; what they give you is a reading taken the same way every time.