How to Track Brand Mentions in AI Search

To track brand mentions in AI search, ask AI tools the questions your buyers ask, on a fixed schedule, and log whether each answer names you, cites you and describes you correctly. Then add the first-party data that exists: Google's generative AI performance report, Bing's AI Performance report and AI referrals in your analytics. No single source shows everything, so the method is a sample plus reports. You'll need Search Console, Bing Webmaster Tools and analytics access. This guide is part of our AI search optimization guide.

Mentions, citations and referrals: what to track

Three different things happen when an AI tool answers a question about your market, and each needs its own column in your log.

A mention is your business named in the answer text. A citation is a link to one of your pages as a source. A referral is a person clicking through to your site. They don't always travel together. An answer can name you while citing a directory, or cite your guide without naming you.

There's a fourth thing worth recording: accuracy. An answer that names you with the wrong services, outdated details or a muddled description can do more harm than no mention at all. Tracking accuracy is how you find those problems before buyers do.

Where AI visibility data comes from

Some engines now report on AI visibility directly. Others leave you to sample answers yourself. The table below shows what each source can and can't tell you.

Use the reports for trends and the prompt log for detail. Neither replaces the other.

How often to check, and who owns it

Monthly is the right rhythm for most businesses. AI answers shift too much day to day for weekly checks to mean much, and quarterly is too slow to catch a wrong fact before it spreads.

Give the job one owner. The same person, using the same prompt list and the same logging rules, produces numbers you can compare. Block a fixed half-day each month for the prompt runs, the report exports and the review.

Check sooner after big changes, such as a site rebuild, a new service or a change to your robots.txt or firewall rules.

Steps to track AI mentions

1. 01
   Define what counts as a mention, a citation and an accurate answer.

   Write the rules down before you start. For example: a mention is your exact business name in the answer; a citation is any link to your domain; an answer is accurate if your services and location match your site. Clear rules let anyone on the team log results the same way, month after month.

2. 02
   Build a fixed set of buyer prompts.

   List 20 to 30 questions in three groups: brand questions ("what does [your business] do"), service questions ("who offers [service] for [type of client]") and problem questions buyers ask before they know your name. Use real wording from sales calls and emails. Keep the list fixed so months compare fairly.

3. 03
   Run the prompts on each engine on a schedule.

   Check Google with AI Overviews and AI Mode, ChatGPT with search on, and Perplexity. Use a fresh chat for each prompt so earlier messages don't steer the answer. Run the full set in the same week each month. Answers vary between runs, so for your most important prompts, run them twice and record both.

4. 04
   Log every result the same way.

   One row per prompt, per engine, per run. Record the date, the engine, the prompt, whether you're mentioned, the cited URL if any, whether the description is accurate and which other sources appear. Note any wrong facts word for word. A plain spreadsheet works. The value comes from consistency, not software.

5. 05
   Read Google's generative AI performance report.

   In Search Console, this report shows impressions of your links in AI Overviews and AI Mode, by page, country, device and date. Export it monthly. Pages gaining impressions show where Google already uses you. Clicks from overviews sit in the main Performance report under the Web search type, mixed with other results.

6. 06
   Read Bing's AI Performance report.

   Bing Webmaster Tools reports how often your pages are cited in Microsoft Copilot, Bing's AI summaries and some partner integrations. It shows cited pages and the grounding queries behind them. Bing says the data is a sample, not a complete log, so use it for trends. If you haven't verified your site in Bing Webmaster Tools, do that first.

7. 07
   Segment AI referrals in analytics.

   Build a segment for sessions whose source or referrer is chatgpt.com or perplexity.ai. Treat the totals as a floor. Some AI visits arrive with no referrer and land in Direct. Compare landing pages against your prompt log to see which cited pages earn clicks.

8. 08
   Review monthly and fix what's wrong.

   Put the month's numbers next to last month's: mentions, citations, accuracy, impressions, cited pages and referrals. Then act. Fix wrong facts at the source, on your own pages first, then on profiles and directories. Add a clear page for any question where you're never mentioned. Recheck those prompts next month.

To choose which of these numbers belong in a monthly report, see AI visibility metrics and KPIs. For a ready-made monthly layout, see our AI search reporting template.

Tracking this every month takes real time. Our in-house SEO team does it for clients end to end, including reporting on where they're mentioned in Google AI Overviews and ChatGPT.

Troubleshooting

Results change every time you check

Cause: this is normal. AI answers are generated fresh each time, and ChatGPT and Google both run related searches behind the scenes, so the sources can shift between runs. Google also says AI Overviews and AI Mode may use different models, so the two won't match each other either.

Fix: don't judge from one run. Keep prompts fixed, use a fresh chat each time, and run key prompts more than once per check. Report rates rather than single results, such as "mentioned in X of Y runs this month." Look at three months of trend before deciding a change worked. Use the Search Console and Bing reports to confirm the direction your sample suggests.

AI answers describe your business incorrectly

Cause: the engine is repeating what it found. That may be an old version of your own page, a directory listing with outdated details, or an article that got your services wrong. If your site never states the facts plainly, other sources fill the gap.

Fix: note the wrong fact and, where shown, the cited source. Update your own pages first so they state the correct facts in plain text near the top. Then correct the profiles and directories that repeat the error, and ask publishers of older articles for an update. Check crawler access so engines can read your fixes, as covered in how to get cited by ChatGPT. Recheck the same prompts next month.

Analytics shows almost no ChatGPT traffic

Cause: there may be little traffic yet, or it may be hiding. Some AI visits arrive with no referrer and get filed as Direct, so the visits you can see are a floor, not the full total. Small sites may also see too few visits for monthly trends to mean much.

Fix: check that your segment matches the source or referrer chatgpt.com exactly, not a misspelling. Look at Direct traffic to pages your prompt log shows ChatGPT cites. A rise there can point to untagged AI visits. Then check your prompt log. If ChatGPT rarely cites you, the fix is visibility work, not analytics settings.