AI search visibility is how often AI answer engines such as ChatGPT and Gemini cite or mention your website when they answer the questions your customers ask. You measure it by running a fixed set of prompts on a schedule, recording the source URLs each answer cites, and calculating how many of those answers include your domain.
Key takeaways
- AI answers vary from run to run, so measure visibility as a rate across repeated checks, never as a single "AI rank".
- Track cited source URLs, not just brand names in the text: citations can be verified, can send clicks and show which pages win.
- Two numbers cover most needs: citation rate (the share of answers that cite you) and share of citations (your share of all cited sources).
- Keep the conditions fixed: same prompts, same engines, same country, clean sessions, a weekly schedule.
- Search Console, Bing Webmaster Tools and analytics give partial first-party data; none of them shows ChatGPT or Gemini citations prompt by prompt.
What is AI search visibility?
AI search visibility measures how present your brand and site are in AI-generated answers from AI platforms such as ChatGPT search, Gemini, Google's AI Overviews and AI Mode, Microsoft Copilot and Perplexity. It is the AI counterpart of search engine visibility, which measures how prominently your site shows up in classic search results for the keywords you care about, and it is the number that GEO and AEO work tries to raise (see how GEO, SEO and AEO differ).
The two need different tracking: an AI answer isn't a ranked list of ten links but a written answer built from a few sources that change between runs.
Google rank tracking | AI visibility tracking | |
|---|---|---|
What you check | A keyword | A prompt or keyword sent to an AI engine with web search on |
What comes back | An ordered list of results | A written answer with a few cited sources |
Result of one check | A position | Cited or not cited, plus the list of cited URLs |
Stability | Fairly stable from day to day | Changes from run to run |
Metric to watch | Position and its change | Citation rate and share of citations across many checks |
First-party data | Google Search Console | Partial only (see below) |
The sources also depend on searches you never see. OpenAI says ChatGPT search rewrites a prompt into one or more targeted queries for its search providers. Google's documentation for Grounding with Google Search describes the same pattern for Gemini models: the model decides whether a search would improve the answer, then "automatically generates one or multiple search queries and executes them". This is why your Google position for one keyword doesn't predict whether an AI answer cites you.
Mentions vs citations: what should you track?
An AI answer can include your brand in three ways, and they are not equally useful to track:
- It cites your page. Your URL appears as a source. This is verifiable, can send clicks, and tells you exactly which page earned the citation.
- It names your brand in the text without linking to you. That helps awareness, but it is harder to check at scale and sends no traffic.
- It cites a third-party page that mentions you, such as a review or a "best tools" roundup. Your brand gets in through someone else's page.
Make citations of your own domain the core metric, because they're objective: a cited URL either is on your domain or it isn't. Then read the cited URLs to find the competitor pages and roundups that shape answers in your category.
Don't track your position inside the answer. In SparkToro's January 2026 research, 600 volunteers ran 12 prompts through ChatGPT, Claude and Google's AI a combined 2,961 times. There was a less than 1 in 100 chance that ChatGPT or Google's AI would return the same list of brands in any two responses, and the same order came up closer to once in 1,000 runs. The author concluded that "visibility % across dozens to hundreds of prompts run multiple times is a reasonable metric", while a "ranking position in AI" is not.
How to track AI search visibility: the method
The method below works whether you track AI mentions by hand in a spreadsheet or with an AI visibility tool that records cited sources.
- Build a prompt set. Start from the keywords you already track and the questions prospects ask in sales calls. Aim for 20 to 50 prompts, mostly without your brand name: category prompts ("best CRM for a small law firm"), problem prompts ("how to automate client intake"), comparisons ("[your product] vs [competitor]") and a few branded ones. Group them by topic so you can see where you're weak.
- Keep the conditions fixed. Use the same engines with web search on, the same country and the same kind of session every time. OpenAI says ChatGPT search uses an approximate location from the user's IP address and that saved memories can shape how it rewrites a query, so use logged-out or clean sessions.
- Run each prompt more than once, or run many prompts. One answer is an anecdote. Totals across many runs are the signal: in SparkToro's follow-up test of 142 headphone prompts, the leading brands appeared in 55% to 77% of 994 responses, a pattern stable enough to track.
- Record the cited source URLs. For every answer, log the engine, the prompt, the date, each cited URL in order, whether any URL is on your domain or a subdomain, and which competitor domains appear.
- Calculate the metrics in the table below.
- Repeat weekly and compare averages. Look at four-week averages rather than single weeks, and note page changes so you can connect cause and effect.
Metric | How to calculate it | What it tells you |
|---|---|---|
Citation rate | Answers that cite your domain ÷ all answers | How often you're a source at all |
Share of citations | Citations of your domain ÷ all citations across all answers | Your share of the sources AI engines rely on |
Competitor share | The same two numbers for each competitor domain | Who wins the answers you want |
Prompt coverage | Prompts where you were cited at least once ÷ all prompts | Which topics you're missing entirely |
Cited pages | Count of citations per URL on your site | Which of your pages earn citations |
Share of citations is close to the Citation Share metric Microsoft added to Bing Webmaster Tools in June 2026, which Bing calculates per grounding query as "the percentage of citations attributed to your site out of all citations shown across all sites for that same grounding query".
Worked example: calculating citation rate and share
The numbers below are an illustrative example, not client data.
A legal-software company tracks 25 prompts in ChatGPT and Gemini, so each weekly check produces 50 answers. Across those answers, the engines cite 380 source URLs in total.
Domain | Answers citing it (of 50) | Citation rate | Citations (of 380) | Share of citations |
|---|---|---|---|---|
Your site | 9 (ChatGPT 3, Gemini 6) | 18% | 11 | 2.9% |
Competitor A | 21 | 42% | 34 | 8.9% |
A legal-tech review site | 14 | 28% | 16 | 4.2% |
The log shows where the gap comes from. Competitor A is cited mostly for comparison prompts, through its "vs" pages, and the review site appears in 14 answers without listing your product. So: publish comparison pages for the matchups prospects ask about, and get reviewed on the site the engines already trust. If the citation rate climbs from 18% toward 30% in the next month's averages, the work is paying off.
Where to get first-party AI visibility data
Some platforms report on AI features directly, but each covers only part of the picture:
- Google Search Console. Google says sites appearing in AI features such as AI Overviews and AI Mode are included in the overall search traffic in the Performance report, within the "Web" search type, so that traffic is counted in your normal totals.
- Bing Webmaster Tools. The AI Performance report, in public preview since February 10, 2026, shows citations of your content across Microsoft Copilot, AI-generated summaries in Bing and select partner integrations, with total citations, cited pages and grounding queries (the grounding queries are a sample). Bing says the Citation Share metric it added in June is "not a ranking system or a competitive scoreboard" and doesn't expose competitor domains.
- Analytics referrals. Visits from chatgpt.com and other AI assistants show up as referrals, but not reliably: on October 14, 2026, Search Engine Roundtable reported, citing an analysis by Adobe's Cedric Huesler, that links opened from GPT-6 answers were arriving without the
utm_source=chatgpt.comtag or a referrer, so analytics filed them as direct traffic. OpenAI had not confirmed the change in that report. - Server logs. OAI-SearchBot and ChatGPT-User requests show your pages are fetched, not cited.
None of these sources tells you, prompt by prompt, whether ChatGPT or Gemini cited you. That gap is what prompt-based tracking fills.
What to look for in an AI visibility tool
A good AI visibility tool records what you would log by hand, on a fixed schedule, so you can compare like with like. Check these points before you choose one:
Criterion | Why it matters |
|---|---|
Engines checked with web search on | ChatGPT and Gemini each run their own searches, so one engine's result says little about the other |
Cited source URLs, not just yes or no | URLs show which of your pages earned the citation and which competitor pages and roundups win |
A country per prompt or keyword | ChatGPT uses an approximate location from the user's IP address, so answers can differ by market |
Where the prompts come from | Writing your own prompts gives control; reusing tracked keywords saves setup and keeps AI and Google data side by side |
Check frequency | Weekly suits most sites; more frequent checks mainly add run-to-run noise unless you average them |
History and export | Trends need stored results; if a tool keeps only the latest check, log it yourself each week |
What the price is based on | Count the prompts or keywords, engines and countries you need, then price that setup |
How RankTera tracks ChatGPT and Gemini visibility
RankTera uses the keywords you already track as the queries. Once a week, it asks ChatGPT (OpenAI) and Gemini to search the web for each tracked keyword and records whether your domain is among the sources they cite: Mentioned or Not mentioned, per engine, in the same table as your Google position. Open a keyword's SERP Details to see up to 10 cited source URLs per engine, with your page highlighted. AI visibility is included for every tracked keyword on every Rank plan, with no add-on.
To apply the method above, count the keywords marked Mentioned and divide by the keywords checked: that is your keyword-level citation rate for each engine. Then read the cited URLs to see which competitor pages and roundups are winning.
Be clear about what the check measures. It records citations, meaning your URLs among the cited sources, not brand names in the answer text, and it doesn't score sentiment. The table shows the latest weekly result for each keyword, so if you want a trend over time, note your Mentioned count each week in a spreadsheet.
What we found: how the top pages talk about AI visibility
When we ran "ai search visibility" through RankTera's KW Research (US Google results, October 15, 2026), it analysed five top-ranking pages from searchable.com, seranking.com, ahrefs.com, medium.com and rankscale.ai. They averaged 1,288 words, and RankTera extracted 79 keywords from the text under their headings. The most used were "AI visibility" (6 times per page on average), "AI search" (4), and "AI Mode" and "brand mentions" (3 each), followed by a group at 2 each that includes "AI Overviews", "AI answers" and "AI platforms".
Our SERP analysis of six top-ranking Google results on the same day found a median length of 1,886 words; three of the six had an FAQ section and three used tables. Three of the six were tool comparisons or tool lists, and one was an agency's service page. The shared vocabulary is brand mentions and Google's AI features; this guide puts the weight on the measurement method and the calculation.
Common mistakes in AI visibility tracking
- Checking once, by hand, while logged in. Memory and location change the answer; one personalized answer tells you little.
- Tracking only branded prompts. Buyers' discovery prompts don't name you, so track those.
- Treating answer order as a rank. Order inside an answer is close to random between runs.
- Mixing countries. Keep each prompt tied to one market.
- Counting brand names without checking sources. A name in the text with no link is weaker evidence than a cited URL.
- Reacting to one week. Compare multi-week averages before you change your strategy.
How to improve your AI search visibility
To earn more citations, rank in classic search for the sub-questions behind each prompt, publish specific, original pages and get mentioned on other sites: in an Ahrefs study of 75,000 brands, branded web mentions correlated with AI visibility far more strongly than Domain Rating did. Our guides to ChatGPT SEO and ranking in Google AI Overviews cover each engine, and the generative engine optimization guide covers the full strategy. Keep watching your Google positions for those sub-questions with a rank tracker, since they feed what AI engines retrieve.
Frequently asked questions
What is an AI visibility tool?
An AI visibility tool sends your prompts or keywords to AI engines such as ChatGPT and Gemini on a schedule and records whether your site or brand appears in the answers. Useful tools log the cited source URLs, not just a yes or no, so you can see which pages earn citations and which competitors win the answers you want.
How do I track brand mentions in ChatGPT?
Pick 20 to 50 prompts your buyers would ask, run them in ChatGPT with search on in a logged-out session, and record the cited source URLs and whether your brand appears. Repeat weekly and calculate the share of answers that cite you, or use a tool that runs the checks and stores the cited URLs.
How often should I check AI search visibility?
Weekly suits most sites. AI answers change from run to run, so daily checks mostly add noise, while monthly checks react slowly after you publish or update pages. Whatever the cadence, compare averages over several weeks and keep the prompts and countries the same between checks.
Does Google Search Console show AI Overviews traffic?
Partly. Google says sites appearing in AI features such as AI Overviews and AI Mode are included in the overall search traffic in Search Console's Performance report, within the "Web" search type. Since August 31, 2026, the Generative AI performance report also shows those AI-feature impressions by page, country, device and date (not by query). For ChatGPT and Gemini citations, you need prompt-based tracking.
Is AI search visibility the same as search engine visibility?
No. Search engine visibility measures how prominently your pages appear in classic search results for your keywords, based on rankings. AI search visibility measures how often AI engines cite or mention you in generated answers. They are linked, because AI engines retrieve pages from search results, but a good Google position for one keyword doesn't guarantee citations.
See your Google rankings and AI citations in one table
AI search visibility is easiest to manage next to the rankings that feed it. RankTera puts both in one table: your Google position for every tracked keyword and a weekly ChatGPT and Gemini check that shows whether your domain is cited, with the cited URLs. Start your 14-day free tier — no card needed: Rank Starter with 300 keywords checked weekly, and nothing is charged when it ends. Plans are on the RankTera pricing page.