AI visibility is how often, how prominently and how accurately your brand shows up when people ask AI assistants like ChatGPT, Google’s AI Overviews or Microsoft Copilot for answers. This guide covers what it means, which metrics matter, which first-party data now exists, and how to measure and report it. For definitions of GEO, AEO and related terms, see our AI visibility, GEO and AEO glossary.
What is AI visibility?
AI visibility (also called AI search visibility or LLM visibility) is your brand’s presence in the answers that AI assistants and AI search features generate. SEO asks, “Where do we rank on Google?” AI visibility asks, “When a buyer asks AI for options in our category, does it name us, cite us and recommend us?”
Mentions, citations and recommendations
- Mention: the answer includes your brand name, whether in a list, a comparison or an example.
- Citation: the answer links to one of your pages as a source. You can be mentioned without being cited, and cited without being mentioned by name.
- Recommendation: the AI suggests you as a fit for the user’s need (“for a 50-person company, consider…”). It’s the most valuable outcome and the most dependent on how the question is asked.
Measure them separately. A brand can get plenty of mentions while the AI pulls its facts from third-party sites, or get cited as a technical source without ever being recommended.
AI visibility vs. SEO rankings
AI visibility doesn’t replace SEO; it sits on top of it. SEO gets your pages into results. AI visibility measures what happens to your brand inside the generated answer. For how GEO, AEO and SEO compare, see the glossary.
Why AI visibility is now a reporting line, not a side project
AI answers now sit in front of a large share of buyer research. ChatGPT has 1.2 billion weekly users, according to OpenAI (November 2026), and Google says AI Overviews has more than 2.5 billion monthly active users. In the US, 44% of adults say they use ChatGPT (Pew Research Center, 2026). And when Google shows an AI summary, people click a traditional result less often: in Pew’s analysis of March 2025 browsing data, users clicked a result on 8% of visits with an AI summary versus 15% without one.
That means a growing part of your brand’s discovery happens in answers your analytics never sees. We cover the full case in why AI visibility matters now.
The 8 AI engines to track
Each engine answers differently and shares different data with brands. Here’s what first-party data exists for each (as of November 2026):
| Engine | How your brand shows up | First-party data for brands | Infinita guide |
|---|---|---|---|
| ChatGPT | Answers with or without sources; business profiles | Business Hub (beta): aggregated mentions and website clicks. Referral links carry utm_source=chatgpt.com |
How to rank in ChatGPT |
| Google AI (AI Overviews and AI Mode) | Summary above the results, or a conversation in AI Mode, with links | Search Console Generative AI performance report: impressions by page, country, device and date | How to get cited in Google AI Overviews and AI Mode |
| Microsoft Copilot | Answers with citations | AI Performance in Bing Webmaster Tools (public preview): citations in Copilot and Bing AI summaries | No guide yet |
| Perplexity | Answers with numbered citations | We found no first-party report for brands; measured by sampling | How Perplexity picks its sources |
| Gemini (app) | Conversational answers | We found no first-party report for brands (Search Console covers Google Search, not the Gemini app) | No guide yet |
| Claude | Conversational answers | We found no first-party report for brands | No guide yet |
| Grok | Answers on X, grok.com and the Grok apps | We found no first-party report for brands | No guide yet |
| DeepSeek | Conversational answers | We found no first-party report for brands | No guide yet |
“We found no” means exactly that: as of this guide, we didn’t find an official brand report for that engine. If one launches, we’ll update the table.
AI visibility metrics that matter
There’s no single standard, but these six metrics cover what leadership usually asks. Calculate each one over a fixed set of prompts that reflects how your buyers search.
| Metric | What it tells you | How to calculate it |
|---|---|---|
| Mention rate | How often AI names you | Answers that mention your brand ÷ total answers |
| Citation rate | How often AI uses your pages as a source | Answers that link to your site ÷ total answers |
| AI share of voice | Your weight versus competitors | Your mentions ÷ mentions of all brands in your category |
| Prominence | Whether you show up early, mid-answer or last | Average position across the answers where you appear |
| Accuracy and sentiment | Whether what AI says about you is correct and favorable | Review of each answer: correct facts; positive, neutral or negative tone |
| AI referral traffic | Visits from AI engines to your site | Sessions with an AI source in your analytics |
Mention rate
The baseline. If your brand appears in 30 of 100 answers to “best payroll software for small businesses,” your mention rate is 30%. Break it down by engine and by prompt type.
Citation rate
Shows whether AI uses your content as a source. If no one cites you, AI is learning about you through other sites. The full definition is in the glossary.
AI share of voice
Compares your presence with competitors in the same answers, so you can tell whether you’re gaining ground even when total category mentions move up or down.
Prominence (and why “AI rank” is unstable)
AI lists reshuffle constantly. In SparkToro’s study, the odds of seeing two lists in the same order were roughly 1 in 1,000. A single “#2 in ChatGPT” screenshot proves very little; average your position across many answers instead.
Accuracy and sentiment
Showing up with an outdated price, a discontinued product or the wrong positioning can hurt more than not showing up. Check what AI says, not just whether it names you.
AI referral traffic
The part you already see. OpenAI adds utm_source=chatgpt.com to referral links from ChatGPT, so those visits are identifiable in Google Analytics. But traffic only counts people who click. Many buyers read the answer and decide without visiting, so referral traffic always understates your real visibility.
A note on “AI visibility scores”
Many tools roll several metrics into one composite score. That’s useful for tracking a trend inside one tool, but each vendor weighs things differently, so scores aren’t comparable across tools. When you report a score, also report the underlying metrics.
How to measure AI visibility
- Build your prompt set. Write the questions your buyers ask, both branded (“Is [your brand] good for remote teams?”) and non-branded (“What payroll software do you recommend for a 50-person company in Texas?”). Non-branded prompts show whether AI picks you when nobody names you. If you’re not sure how buyers phrase things, start with how your customers actually ask AI.
- Choose your engines. Ideally all 8. If you’re starting by hand, begin with the ones your audience uses most.
- Run each prompt several times. One answer tells you little. SparkToro suggests that to really know an AI’s recommendations you need to ask over and over (60 to 100 times per prompt in their study) and average the results.
- Log the same fields every time: whether you appear, where, which sources are cited, which competitors are named, and whether the facts about you are correct.
- Cross-check with first-party data and repeat on the same prompt set to see the trend.
First-party data you already have
- Google Search Console: the Generative AI performance report, announced in June 2026 and rolled out to all sites on August 31, 2026, shows how often links to your site appeared in AI Overviews and AI Mode, by page, country, device and date. It doesn’t show clicks, queries or position.
- Bing Webmaster Tools: AI Performance, in public preview since February 2026, shows when your site is cited in Microsoft Copilot and Bing’s AI-generated summaries.
- ChatGPT Business Hub: since November 10, 2026 (beta), businesses can claim their profile in ChatGPT and see aggregated insights such as mentions and website clicks. In the US, eligible ecommerce and online service businesses can also launch a Business Agent.
- Google Analytics: sessions from ChatGPT (tagged
utm_source=chatgpt.com) and from other assistants that pass a referrer.
Prompt sampling for everything else
For Gemini, Perplexity, Claude, Grok and DeepSeek, we found no first-party brand report. And even where official data exists, it doesn’t tell you which competitors appear next to you or what the AI says about you. For that, you need to run prompts systematically and keep the answers.
Branded vs. non-branded prompts
Branded prompts tell you what AI knows about you. Non-branded prompts tell you whether AI chooses you before the buyer knows your name. You need both: a brand can look great when named and never appear in its own category.
For the full method, with the prompt set, sample size, tracking sheet and formulas, see How to track AI visibility: a prompt-by-prompt method.
How to report AI visibility to leadership
These are our recommendations, based on what we see teams struggle with:
- Keep AI impressions separate. Search Console’s AI impressions aren’t the same unit as classic organic impressions. Report them on their own line rather than blending them into “total search visibility.”
- Lead with non-branded mention rate and share of voice by engine. That’s the closest thing to “are we in the consideration set?”
- Show accuracy issues as risks, with the exact answer text.
- Report referral traffic as a floor, not the full impact.
- Compare month over month on the same prompt set. Changing the prompts resets the baseline.
- Note what you changed. Content updates, PR wins and new reviews, so trends have context.
What drives AI visibility
No engine publishes the formula, but two signals have solid evidence behind them:
- What the web says about you. In an Ahrefs study of 75,000 brands, branded web mentions had the strongest correlation (0.664) with appearing in Google’s AI Overviews, far above backlinks (0.218). It’s a correlation, not proof of cause, but it points to press, reviews, communities and comparison content.
- Content that’s easy to cite. In the academic paper that coined GEO (KDD 2024), adding sources, quotations and statistics improved a page’s visibility in generative answers by up to 40%, while keyword stuffing did worse. It was a benchmark experiment, not a guarantee for your site.
Engine-specific tactics are in our guides to ChatGPT, Google AI Overviews and Perplexity.
Example: a monthly AI visibility snapshot
Illustrative example with a fictional brand and hypothetical numbers; not real data.
“ExampleCo” is a fictional payroll software company in Austin. Its team tracks 20 prompts, 5 branded and 15 non-branded, and runs each 5 times in ChatGPT, Google AI and Microsoft Copilot: 300 answers in total.
| Metric | ChatGPT | Google AI | Copilot |
|---|---|---|---|
| Mention rate (non-branded prompts) | 16% | 24% | 8% |
| Citation rate (all prompts) | 4% | 18% | 12% |
| Inaccurate facts (branded prompts) | Old pricing tier | None | None |
What the team reports: strongest in Google AI, likely because the site already ranks well in Google; weak in ChatGPT, which also repeats outdated pricing; cited by Copilot for its help docs but rarely recommended. Next month’s plan: fix the outdated pricing at the source, earn placements in independent payroll comparisons, and re-measure on the same 20 prompts.
ChatGPT
Weakest, repeats old pricing
- Mention ratenon-branded prompts16%
- Citation rateall prompts4%
Inaccurate facts
Old pricing tier
Google AI
Strongest
- Mention ratenon-branded prompts24%
- Citation rateall prompts18%
Inaccurate facts
None
Copilot
Cited, rarely recommended
- Mention ratenon-branded prompts8%
- Citation rateall prompts12%
Inaccurate facts
None
Next month
- 1Fix the outdated pricing at the source
- 2Earn placements in independent payroll comparisons
- 3Re-measure on the same 20 prompts
If your brand isn’t showing up
Not appearing usually has specific causes: AI doesn’t know your brand, knows it but doesn’t connect it to your category, or has outdated information. How to check each one: Why isn’t my brand showing up in ChatGPT? What to do.
Guides in this topic
- How to rank in ChatGPT: a practical guide for brands
- How to get cited in Google AI Overviews and AI Mode
- How Perplexity picks its sources, and how to get it to cite you
- Why isn’t my brand showing up in ChatGPT? What to do
- How your customers actually ask AI (and how to get recommended)
- Why AI visibility matters now, not “someday”
- GEO, AEO and AI visibility: a practical glossary for marketing teams
How Infinita helps
Measuring by hand works to start, but it gets slow as soon as you add prompts and engines. Infinita runs your tracked prompts on a schedule in ChatGPT, Gemini, Google AI, Perplexity, Claude, Grok, Microsoft Copilot and DeepSeek, and shows you the prompts where your brand appears and the ones where it doesn’t, for both branded and non-branded prompts. It saves every answer with its sources, so you can see which competitors show up when you don’t and which pages AI relies on to describe your category.
With Ask Vera, your AI marketing agent, you can ask “Which non-branded prompts did we lose this month?” and get a plan to win them back.
Infinita works with teams in technology, retail and ecommerce and other industries. To see where your brand stands in AI answers today, book a demo.