Knowledge ·
What is AI visibility, and how do you measure it?
In short
AI visibility is the measurable presence of a company in the answers of AI assistants: how often it is named or recommended, how it is described and which sources the AI relies on. It is measured not as a rank but as a share of many answers across several questions, engines and days, with a margin of error.
What does AI visibility mean?
AI visibility describes how often, and in what way, a company appears in the answers of AI assistants such as ChatGPT, Gemini, Google AI Overviews and Claude. It is not a position in a list of results but a share: out of all the answers to the questions your customers typically ask, how many include your company, and in what role?
Three levels matter:
- Named: your company appears in the answer, for example in a list.
- Recommended: the answer explicitly advises the reader to choose you.
- Cited as a source: the AI draws on your pages and links to them.
How you are described matters too. An answer can name you and still get you wrong, with an out-of-date offer, the wrong location or a caveat attached.
Where does an AI get its answers from?
AI assistants answer from two places: what a model learnt during training, and a web search it runs while it writes the answer. What a model knows without searching only changes when a new model is released. Pages that a search finds, on the other hand, can often show up in answers within days or weeks.
When they search, the services often split one question into several smaller searches. Google calls this a ‘query fan-out’ for AI Overviews and AI Mode: several related searches across subtopics and data sources, according to Google Search Central (2025). For your company, that means your website has to answer those smaller questions, and other sources have to confirm what you say about yourself.
Why does AI visibility matter for companies?
AI visibility matters because an AI answer names only a handful of firms. If you are missing from the answer to a question such as ‘Which accountant would you recommend for SMEs in Zurich?’, you are not on that person’s list, however well your website does in Google search.
The gap is large. A study of 98 Swiss German SMEs, reported by Netzwoche on 29 September 2026, found that 63% of the firms did not appear in a single AI answer. Across all 2,178 answers analysed, the systems named the firm in question in only 3% of cases.
Yet few companies look. According to the KMU Digital Pulse 2026 survey by the Lucerne University of Applied Sciences and Arts and localsearch, 58% of the 504 SMEs surveyed have never checked how they appear in AI answers. Only 2% have taken targeted action.
AI visibility and SEO compared
AI visibility and SEO share the same foundations but measure different things. Google’s guide (2026) says that, from Google Search’s point of view, optimising for generative AI search is still SEO, and that no new files or special markup are needed for it. ChatGPT and Claude, however, use their own search systems and sources, and the measurement differs for every service:
| Classic SEO | AI visibility | |
|---|---|---|
| Goal | Your page ranks high and gets clicked | Your company appears in the answer, described correctly, ideally recommended |
| What people see | A list of links | A text with a few names and some sources |
| Metric | Rank, impressions, clicks | Share of answers that name or recommend you, with a margin of error |
| Variation | A rank can be read off for a search | The same question does not get the same answer every time |
| What counts | Your page and the links pointing to it | Your page, plus directories, associations, media and reviews |
| How it is measured | Search Console and rank tracking tools | Repeated questions to several AI services |
How the terms GEO, LLMO and AEO fit in is explained in What is LLMO?.
Which metrics make up AI visibility?
AI visibility breaks down into four metrics, each a share of many answers:
- Mention rate: in how many answers does your company appear?
- Recommendation rate: in how many answers does the AI advise choosing you, rather than just listing you?
- Citation share: how often does the answer cite your own pages?
- Share against competitors: how often are you named compared with the other firms that come up for the same questions?
On top of that, check the description itself: are your offer, location and prices right?
How do you measure AI visibility?
You measure AI visibility by asking several AI services the same customer questions several times, on several days, and counting how many answers name the company. A single answer is never a result.
Why one answer is not a result
The same question does not get the same answer every time. Language models choose their words with an element of chance, web search returns different pages on different days, and the engines are updated all the time. If you ask once and are named, all you know is that it can happen. Only a count across many answers shows how often it does.
What a margin of error is
A margin of error shows the range in which the true share probably lies. A 95% margin of error is calculated so that it contains the true value in 95 cases out of 100. The more answers you have, the narrower it gets. A worked example, not a measurement (Wilson method):
| Answers that name you | Share | 95% margin of error |
|---|---|---|
| 3 of 10 | 30% | 11 to 60% |
| 30 of 100 | 30% | 22 to 40% |
| 300 of 1,000 | 30% | 27 to 33% |
The share is the same each time, but after 10 answers almost anything between rare and frequent is possible. The figure only becomes useful with hundreds of answers.
When a change is real
A change only counts once it is larger than the margin of error. A worked example: if your share rises from 22% to 25% with a margin of ±6 points, that may be chance. If it rises to 31%, the difference is larger than the margin, so it is real.
How anewera measures
- Several answers per question, engine and day, because answers vary.
- Four engines: ChatGPT, Gemini, Google AI Overviews and Claude.
- A Swiss location on every engine, plus the city on Claude.
- Failed answers are kept and reported. Nothing is smoothed or filled in.
- Every share comes with its margin of error and the number of answers behind it.
The details are on the method page. What this looks like from zero is shown in our live case study, ‘Grow in Public’: since 6 October 2026 we have been measuring our own visibility there. There are no results yet. We only show figures once there are three days of measurements, and always with a margin of error.
Check it yourself in four steps
- Write down 10 to 20 questions the way your customers ask them, with place and language.
- Ask each question several times in each service, on several days, each time in a new chat.
- For every answer, note whether you are named, recommended, cited as a source or missing, and which pages are cited.
- Count, rather than reading individual answers.
That is enough for a first impression. Tracking change over months takes many answers a day, and that takes software. Where your company stands today is what the AI visibility audit measures.
What is different about AI visibility in Switzerland?
In Switzerland, customers ask in four languages, and Swiss directories play a large part as sources.
Languages. According to Netzwoche (2025), Google has shown AI Overviews in Switzerland since 26 March 2025, in German, French, Italian and English. AI Mode followed on 8 October 2025, in German, French and Italian. A hotel study reported by Netzwoche (2026) shows why languages have to be counted separately: a company can be very present in German answers and barely appear in French ones. So measure in every language your customers use.
Local sources. In the SME study mentioned above (Netzwoche, 2026), local.ch was among the sources cited for 63% of the firms examined, search.ch for 24% and comparis.ch for 23%. For questions about Swiss PR agencies, an analysis by Aclipp (persoenlich.com, 2026) found directories cited in 43% of answers, associations in 18% and trade media in 10%. Consistent listings, with the same name, address and offer as on your website, are therefore part of AI visibility.
Data protection. Measuring needs no personal data: what goes to the engines are typical customer questions, not your customers’ details. If you measure yourself or use a tool, you should still know where the raw answers and analyses are stored. anewera stores them in Zurich.
How do you improve AI visibility?
AI visibility is improved in four places:
- Technology: AI crawlers may read your pages, and neither robots.txt nor a firewall locks them out.
- Content: pages that answer your customers’ follow-up questions, with location, offer and prices in plain sentences.
- Sources: listings and mentions where the AI looks: directories, associations, trade media.
- Measurement: before and after, with a margin of error, so you can tell an effect from chance.
Depending on who you ask, this work is called GEO or LLMO. What lies behind the terms is explained in What is LLMO?.
Frequently asked questions
What is the difference between AI visibility and SEO?
SEO measures where your page sits in a list of search results and how often it is clicked. AI visibility measures in how many answers from AI assistants your company is named, recommended or cited as a source. Both rest on the same foundations. Because an AI answer changes from one run to the next, AI visibility is measured as a share with a margin of error, not as a rank.
How do I find out what ChatGPT says about my company?
Ask ChatGPT the questions your customers ask, each in a new chat, several times and on several days. For every answer, note whether your company is named, recommended or cited as a source, and how it is described. A single answer only shows what is possible. Counting across many answers shows how often it actually happens.
Which tool can I use to measure my company's AI visibility?
There are self-service tools and services that measure, make the changes and report. Whatever you choose, check five things: several answers per question and day, the engines your customers use, a Swiss location and your languages, shares with a margin of error, and failed answers that are reported rather than quietly dropped. That is how anewera measures on ChatGPT, Gemini, Google AI Overviews and Claude.
What does AI visibility mean for local businesses?
AI assistants often answer local questions, such as which electrician to call in Bern, with the help of directories. In a study of 98 Swiss German SMEs reported by Netzwoche in September 2026, local.ch was among the sources cited for 63% of the firms. For a local business, complete and consistent listings on local.ch and search.ch matter as much as its own website.
How often should AI visibility be measured?
Continuously, if you want to see change. Answers vary from day to day and the engines are updated all the time. A one-off measurement shows where you stand. Whether a piece of work had an effect only shows when you compare over weeks, with enough answers that the change is larger than the margin of error.
Sources
- Schweizer KMUs bleiben in KI-Antworten oft unsichtbar, Netzwoche,
- KI-Sichtbarkeit: 58 % der KMU haben nie nachgeschaut, Lucerne University of Applied Sciences and Arts, Institute of Communication and Marketing IKM,
- Team Farner vorn, Spezialisten dicht dahinter, persoenlich.com,
- KI-Assistenten verändern die Sichtbarkeit von Schweizer Firmen, Netzwoche,
- Google bringt KI-Suche in die Schweiz, Netzwoche,
- Google startet neuen KI-Modus bei der Suche in der Schweiz, Netzwoche,
- Optimizing your website for generative AI features on Google Search, Google Search Central,
- AI features and your website, Google Search Central,
Retrieved or checked on the date given.