Many brands do not know how frequently, or even whether, ChatGPT, Gemini, or Google's AI mentions them. Not because it is unknowable, but because it sits outside the analytics they already have. This is the systematic way to measure it.
AI visibility answers the same question a ranking used to, with one difference: instead of a single position on a single engine, you are tracking a citation rate across several. Below is how to measure it accurately, and how it connects to AI share of voice if you already track that.
What is AI visibility?
AI visibility is how often, or whether, a brand or its content appears in answers from ChatGPT, Perplexity, Google AI Overviews, AI Mode, Gemini, and Copilot. It covers being quoted, being named in the answer text, and being described correctly.
No engine hands you a score. You build it by running real questions against each engine and recording the responses. It is measurable, but not automatic.
How to measure AI visibility, step by step
The procedure is the same whether you run it in a spreadsheet or in software. Follow it in order.
1. Build a prompt set from real buyer questions
Start with the questions your buyers actually ask rather than your brand name. Sales-call notes and Search Console queries are the two best sources. Fix the set so it covers the whole journey, from problem awareness through to comparison.
That prompt set is your measuring instrument. If the prompts are vanity queries, the results are vanity results. Comparison and recommendation prompts are where brand mentions actually happen.
2. Choose the engines that matter to you
Focus on the engines your audience actually uses. B2B brands mostly rely on ChatGPT, which already has 800 million weekly active users, alongside Google AI Overviews and AI Mode, then Perplexity, Gemini, and Copilot.
Spend the effort where your customers are. Measuring an engine your audience never touches adds noise, not insight.
3. Run the prompts and record presence
Run every prompt on each engine and record three things: whether you appear at all, whether you were named in the text or given a link, and which competitors appeared alongside you. Log the source domains the engine pulled from as well, because that column is where the fix usually comes from.
Do this systematically rather than once. AI responses are non-deterministic, so a single result is one sample rather than a measurement. The trend across repeated runs is the real data.
4. Turn it into a score
Convert the raw log into a rate you can track. The simplest is presence rate, the percentage of prompts where you appear at all. A stricter one is citation rate, the percentage where you are actually listed as a source.
Share of voice sits on top of both, showing how often you are cited or mentioned relative to competitors on the same prompts. Our share of voice guide covers the formula and the sample sizes it needs to be trustworthy.
5. Benchmark against competitors
A visibility figure on its own means very little. A 20 percent presence rate can be excellent or poor depending on the category. Run the same prompt set for your main competitors and compare.
That comparison is also your content plan. The prompts where rivals appear and you do not are the specific gaps to close, and this is the step where measurement turns into action.
6. Add your first-party signals
Prompt tracking covers every engine, but pair it with the free first-party data. Bing Webmaster Tools AI Performance shows your Copilot and Bing citations. Google Search Console's generative AI report shows AI Overview and AI Mode impressions.
GA4's AI Assistant channel catches referral traffic from engines that pass a referrer, and your server logs show which AI crawlers are actually reaching the site.
Each is partial, but together they corroborate your prompt tests with first-party data you own. Our guide to AI search analytics sets out exactly what each source shows and what it is blind to.
7. Set a cadence and watch the trend
Pick a cadence, weekly or monthly, and repeat the whole procedure on the same prompt set. The trend is the point. Watch whether citation rates rise after you publish, whether a competitor has overtaken you, and whether a model update reshuffled the results.
Measurement without a cadence is an audit. Measurement with a cadence is a program.
What are the common mistakes to avoid?
A few traps make AI visibility numbers misleading. Do not treat a single run as fact, because non-determinism guarantees variance. Do not count only links and ignore mentions, since links inside AI summaries are clicked only about 1 percent of the time and a plain mention still shapes the buyer.
Do not test your brand name instead of buyer queries, which inflates the number on questions nobody needed answered. And do not measure in a vacuum, because a raw rate with no competitor comparison tells you almost nothing.
What to do once you can measure it
Measurement is only the first half. The point of the log from step 3 is that each column implies a different fix, and reading them apart is what stops you spending a quarter on the wrong one.
If the engine cited a competitor's page you do not have an equivalent of, that is a content gap, and the complete guide to answer engine optimization and the GEO playbook cover how to close it. If it cited your page but lifted a thin passage, the fix is formatting, and getting cited by ChatGPT covers the answer-first structure that survives extraction. If it cited a third-party roundup you are absent from, no amount of on-page work helps and the job is off-site.
Then narrow or widen the lens as needed. Tracking brand mentions in ChatGPT is the single-engine version of this method, and our GEO tools comparison covers the software once the prompt set outgrows a spreadsheet, which for most teams is the point where manual measurement quietly stops happening. Unveilr runs the loop on a schedule for that reason.
Frequently Asked Questions
How do I measure AI visibility?
Create a fixed set of real buyer prompts, run them across the major engines, and log whether you were mentioned or cited. From there, derive a presence and citation rate, compare against competitors, cross-check with free first-party data like Bing AI Performance and GA4, and repeat on a schedule to see the trend.
What is a good AI visibility score?
There is no universal benchmark, because it depends entirely on your category and who you are competing against. A 20 percent presence rate can be excellent or poor. Your number only means something next to your competitors' numbers, so measure them on the same prompt set and read your visibility as a share of that field.
How is measuring AI visibility different from tracking rankings?
Rankings measure one engine and reward a click. AI visibility measures many engines and rewards a citation or mention, often with no click at all. It is also non-deterministic, so instead of a stable position you track a rate across repeated runs. With roughly two-thirds of searches now ending click-free, the citation is frequently the whole outcome.
Can I measure AI visibility for free?
Yes, but only in part. It's feasible to use manual prompts in spreadsheets and pair it with free first-party sources: Bing Webmaster Tools AI Performance, Search Console generative AI report, GA4 AI Assistant channel and server logs. The constraint here is scale and cross-engine consistency, as realistically doing this by hand across too many prompts and engines quickly becomes cumbersome.
How often should I measure AI visibility?
Measure on a fixed cadence, weekly or monthly, using the same prompt set each time. Because a single prompt returns different results on different days, an isolated run misleads. Rerunning the same prompts on a schedule is what surfaces the real trend and shows when a competitor pulls ahead. Repeating the same prompts over time shows how citation and share of voice figures move after material is published or updated.
Which AI engines should I measure?
Those that your potential customers utilize. Start with ChatGPT, given its 800 million users, then Google AI Overviews and AI Mode, followed by Perplexity, Gemini, and Copilot depending on your niche. Researching an engine your target market does not use adds trouble without useful insights, so focus your prompt list on the engines your customers actually favor.