Insights/How to measure AI visibility, and what not to measure
How to measure AI visibility, and what not to measure
By Kenny Tang · 2026-10-11 · 5 min read

In short
- AI visibility means how often AI assistants mention your business, cite your pages and send people to your site. No single tool measures all of it, so you combine a few sources.
- Google counts AI Overviews and AI Mode inside the normal Search Console Performance report. Bing Webmaster Tools now has a dedicated AI Performance report that counts citations.
- Do not chase a single "AI ranking", one-off screenshots or vanity scores. Track a fixed set of real questions over time, and connect visibility to enquiries.
What "AI visibility" actually covers
When people say they want to be "in ChatGPT" or "in AI Overviews", they usually mean three different things:
- Mentions. The assistant names your business or brand in its answer.
- Citations. The answer links to one of your pages as a source.
- Visits and enquiries. Someone clicks through and then calls, books or fills in a form.
These are related but not the same. You can be mentioned without being cited, for example when an assistant describes you from a directory listing. You can be cited and get few clicks, because the answer already told the reader enough. A sensible measurement plan looks at all three and does not treat any one of them as the whole story.
Researchers have taken this seriously too. The paper that introduced the term Generative Engine Optimization, accepted at KDD 2024, proposed ways of measuring how visible a source is inside a generated answer and reported that optimisation methods could raise that visibility by up to 40% in its tests. The useful lesson for a business is that visibility inside an answer can be measured and compared, not just guessed.
What Google gives you
Google's guidance on AI features and your website says sites that appear in AI Overviews and AI Mode are reported in the Search Console Performance report, within the "Web" search type. There is no separate AI tab. Google also suggests looking beyond clicks at conversions and time on site in analytics tools.
The Performance report help page explains how these appearances are counted:
- Clicking an external link in an AI Overview or in AI Mode counts as a click.
- Standard impression rules apply.
- An AI Overview takes one position on the results page, and every link inside it is given that same position.
What this means in practice: watch impressions and clicks for the queries and pages you care about, and compare periods. A rise in impressions with flat clicks on question-style queries can be a sign that answers are being shown on the page. Do not read the average position for those queries as precise, because one AI Overview can hold several links at the same position.
What Bing gives you
In February 2026 Microsoft opened a public preview of AI Performance in Bing Webmaster Tools. It covers Microsoft Copilot, AI-generated summaries in Bing and selected partner integrations, and shows:
- total citations of your pages in AI answers over a period;
- average cited pages per day and citation counts for each URL;
- grounding queries, the phrases the AI used when retrieving content it went on to cite.
Bing is careful about what the numbers do not mean. The data is a sample of overall citation activity, and the counts do not indicate placement, ranking, authority or how important a page is. In June 2026 Bing added Intents, Topics, Citation Share and Compare in preview. Citation Share is the percentage of citations attributed to your site, and Bing describes it as an observational metric, not a ranking or a competitive scoreboard.
If your site is not yet verified in Bing Webmaster Tools, this is a quick and worthwhile step.
Measuring ChatGPT and other assistants
Most assistants do not offer site owners a dashboard, so you measure them in two ways.
Check you can be found at all. OpenAI's crawler documentation says OAI-SearchBot is used to surface websites in ChatGPT's search features, and that sites which block it will not be shown in ChatGPT search answers. Before measuring anything, confirm your robots.txt and any firewall or CDN rules are not blocking the crawlers of the assistants you care about.
Run a fixed question panel. Write down 15 to 30 questions your customers really ask, in English and in Chinese. Ask each assistant the same questions on a regular schedule, ideally monthly, and record for each answer whether you were mentioned, whether you were cited, which page was used and which competitors appeared. Our ChatGPT recommendation checklist is a good place to start building the list.
Then look in your analytics for visits referred from AI assistants, such as chatgpt.com, perplexity.ai or copilot.microsoft.com, and connect them to form submissions, calls or bookings.
What not to measure
Some numbers look convincing but lead to poor decisions:
- One screenshot. AI answers vary with wording, language, time and model. A single good or bad answer proves very little.
- A single "AI rank". There is no fixed position in a generated answer, and both Google and Bing warn against reading their AI numbers as rankings.
- Citation counts on their own. A page cited often for an unrelated question does not help sales. Look at which queries lead to citations.
- Traffic alone. Fewer, better-informed visits can still bring more enquiries. Judge AI traffic by what it leads to.
- Opaque scores. Be cautious with any tool that gives one "AI visibility score" without showing which questions, assistants and dates it is based on.
A simple monthly routine
- Export Search Console data for your key pages and question-style queries, and compare with the previous month.
- Review Bing AI Performance: total citations, cited pages and new grounding queries.
- Run your question panel in each assistant, in both languages, and update your tracking sheet.
- Check AI referrals and the enquiries they produced in analytics.
- Pick one or two pages to improve based on what you found, and recheck next month.
The short version
Measure mentions, citations and enquiries, use the reports Google and Bing already give you, and track a fixed set of real questions over time. Ignore single screenshots and any number that claims to be your AI ranking. If you want help setting up this kind of tracking, see our AI marketing and GEO service or contact us.
Frequently Asked Questions
General
Can I see AI Overviews traffic separately in Google Search Console?+
Not as its own report. Google says sites appearing in AI features are counted in the Performance report under the Web search type, together with normal results. An external link clicked in an AI Overview or AI Mode counts as a click, and every link inside one AI Overview shares the same position.
Which tool shows citations in AI answers today?+
Bing Webmaster Tools has an AI Performance report, in preview, that counts how often your pages are cited as sources in Microsoft Copilot, AI summaries in Bing and some partner integrations. Bing notes the data is a sample and that citation counts do not indicate ranking or authority.
Is a single prompt test in ChatGPT a reliable measure?+
No. AI answers change with wording, language, time and the model in use, so one screenshot says little. Use a fixed list of real customer questions, run it on a regular schedule in each language and assistant, and look at the trend over several weeks rather than any one answer.
Sources
- https://developers.google.com/search/docs/appearance/ai-features
- https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview
- https://blogs.bing.com/search/June-2026/New-AI-Visibility-Insights-in-Bing-Webmaster-Tools-Intents-Topics-Citation-Share-Compare
- https://developers.openai.com/api/docs/bots
- https://arxiv.org/abs/2311.09735
