How to measure AI visibility

What does AI visibility mean?

AI visibility is how often, how prominently and how well your brand appears in answers from AI engines. It covers ChatGPT, Perplexity, Gemini, Claude and Google's AI Overviews and AI Mode. It is measured from the answers to questions your buyers actually ask.

There is no public dashboard for it. The engines do not publish who they mention. So AI visibility is always measured by sampling. You ask a fixed set of questions, record the answers and work out the numbers from those.

Which numbers measure AI visibility?

Five numbers do most of the work. Each one answers a different question, so it helps to look at them together rather than rely on one.

Mention rate alone can flatter you. Being named last in a list of eight, with a caveat, is not the same as being named first and recommended. That is why position and quality measures sit alongside it. The share of voice guide shows how position weighting works with a worked example.

How to measure AI visibility step by step

The steps below work whether you measure by hand or with a tool. The method matters more than the software.

1. Fix your prompt set

Choose 20 to 50 real buyer questions for each market. Include research questions, comparisons and supplier questions. Keep the list stable. If it changes every month, you cannot tell whether a change in score came from the engines or the list. The tracking guide covers how to build it.

2. Choose engines and markets

Measure the engines your buyers use. For most UK businesses that means ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. If you sell in several regions, measure each one separately.

3. Run each prompt repeatedly

Run every prompt on a schedule, weekly or fortnightly. Where you can, run each one more than once per period. The more answers you collect, the steadier your numbers will be.

4. Calculate mention rate

Divide the answers that name you by the total answers. Do this per engine as well as overall. A brand can be strong in Perplexity and weak in Gemini, and the overall figure would hide that.

5. Weight by position

Give each mention a weight of one divided by its position. Named first scores 1, second scores 0.5, third scores about 0.33. Add these up for each brand and work out each brand's share of the total. This is position-weighted share of voice.

6. Add quality measures

Record sentiment, whether you were recommended, whether the facts about you were right, and which of your pages were cited. These tell you whether visibility is helping or hurting.

7. Link to outcomes

Visibility only matters if it leads somewhere. Google Search Console's generative AI performance report shows impressions from AI Overviews and AI Mode. Bing Webmaster Tools has an AI Performance report for citations in Copilot. Google Analytics 4 shows visits from AI tools as referrals.

8. Read trends, not snapshots

Compare periods of several weeks. Check that a change holds for more than one period before acting on it. Then look at which prompts moved, because that tells you what caused it.

What is an AI visibility score?

An AI visibility score is a single number that sums up how visible a brand is across a set of questions. Most tools build it from mention rate and position. Some fold in sentiment, recommendations and citations too.

A single score is handy for a chart or a board report. But it hides detail. Two brands with the same score can have very different problems. One may be named often but always last. The other may be named rarely but always first and recommended. Always check the parts behind the score.

What is a good AI visibility score?

There is no universal good score. Each tool calculates its score in its own way, and each prompt set is different. A score of 40 in one tool may mean something quite different from 40 in another.

A more useful test has three parts. Are you ahead of the competitors you lose work to, on the same questions? Is your score rising over several periods? And is it rising on the questions that matter most, those closest to a buying decision? If the answer to all three is yes, your score is good, whatever the number is.

Be wary of any benchmark that claims a fixed good score for every business. A national brand and a local firm with one branch should not be judged against the same figure.

Why do AI visibility numbers move from week to week?

AI answers vary from run to run, so every visibility number is a sample. Language models add some randomness when they write. Search-enabled engines also pick up new pages, and consumer apps can use location and history. OpenAI says ChatGPT may use location and saved memories when it rewrites a question into searches.

The size of the sample makes a big difference. Say you track 40 questions, three runs each, which gives 120 answers. You are named in 36 of them, a mention rate of 30%. Basic sampling maths says the true rate could easily sit anywhere from about 22% to 38%. So a move from 30% to 33% next week may mean nothing at all.

The same maths shows how to get steadier numbers. Track more questions, run them more often, and compare longer periods. Treat small moves as noise until they hold.

How do you connect AI visibility to business results?

AI visibility is a leading measure. It shows whether buyers are likely to hear about you. The results you care about, enquiries and sales, come later and are harder to pin to one cause.

A few links are worth building. Google says clicks from its AI features are included in the normal Search Console performance data, within the Web search type. So watch clicks and conversions on the pages that AI answers cite. In Google Analytics 4, track visits from AI tools and what those visitors do. On enquiry forms and sales calls, ask how people found you, and add "an AI tool such as ChatGPT" as an option.

None of these gives a perfect line from mention to sale. Together they show whether rising visibility is followed by more visits and more enquiries. When you report to a board or a client, show the visibility trend next to these outcome figures. Add a plain note on what each one can and cannot prove. That is also where clear reports earn their keep.

How accurate are AI visibility tools?

AI visibility tools are as accurate as their sampling and their reading of answers. Three things are worth checking before you trust one.

How answers are collected. Tools that use the engines' APIs get consistent, repeatable answers. Those answers can differ a little from what a person sees in the app, because personal history and location are not included. That is a fair trade for consistency, as long as the tool says so.

How answers are read. A tool must spot your brand name, variants and misspellings, work out position and judge tone. Ask how it handles brands with common words in their names.

How many answers sit behind each number. A score from ten answers is fragile. A score from several hundred is much steadier. The guide to choosing an AI visibility tool covers what else to ask.

How Axiom GEO helps

AI visibility tracking in Axiom GEO runs your prompts on a schedule across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. It reads each answer for mention, position, sentiment, competitors, recommendation and cited pages. It reports share of voice weighted by position, using 1/rank, and a 0 to 100 prominence score that folds in recommendation, citations and sentiment. Answer accuracy, on the Professional plan and above, tracks an accuracy score over time and per engine. AI traffic analytics joins Google Analytics 4 and Search Console data to show visits from AI tools.