How to Track Whether AI Search Mentions Your Business

More people now ask an AI assistant the questions they used to type into a search box: which med spa nearby does lip filler, what a dental implant costs in their city, which protein powder has third-party testing. For a local practice or a small brand, the useful question is not whether "AI SEO" is the future. It is simpler: when someone asks, does your business come up, is it described correctly, and does anyone click through?

This guide is about measuring that. It covers the free data the platforms now give you, how to run your own prompt tracking, what to record, and where paid tools fit. For the content and site changes that tend to improve visibility, see our companion GEO checklist for content and site structure.

Know which engines you are measuring

Each AI answer engine gets its information from a different place, so each needs its own check.

Google AI Overviews and AI Mode draw on Google's own search index and ranking systems. Google's guidance says that if a page is indexed and eligible to show with a snippet in regular Search, it is eligible for these features.
ChatGPT search finds pages through OpenAI's search crawler, OAI-SearchBot. OpenAI says sites that block it will not appear in ChatGPT search answers, apart from plain navigational links.
Perplexity uses its own crawler, PerplexityBot, to surface and link sites in its answers.
Microsoft Copilot and the AI summaries in Bing are grounded in Bing's index.

Being strong in Google does not mean you show up in ChatGPT or Copilot, and the reverse is also true. A measurement plan should cover at least Google, ChatGPT, Perplexity and Copilot, plus Gemini if your customers use it.

Start with the free first-party data

Three sources of data cost nothing and come straight from the platforms. Set them up before paying for anything.

Google Search Console: Generative AI performance report. Google introduced this report in June 2026 and says it reached all websites worldwide by August 31, 2026. It shows impressions, meaning how often links to your site appeared in AI Overviews, AI Mode and AI features in Discover. You can group by page, country, date and device. It does not show clicks (those stay in the regular Performance report), and it does not separate AI Overviews from AI Mode. Sites with very few AI impressions may not see the report at all.

Bing Webmaster Tools: AI Performance. Microsoft launched this as a public preview in February 2026. It counts how often your pages are cited in Copilot, in AI-generated summaries in Bing and in some partner integrations. It also shows the "grounding queries" the AI used when it pulled your content, and page-level citation counts. In June 2026 Bing added intent, topic and citation share views and a compare option. Microsoft is clear that a citation is not a click or a visit.

Google Analytics 4: AI referral traffic. When someone clicks a link in an AI answer, the visit often shows up as a referral from chatgpt.com, perplexity.ai, copilot.microsoft.com or gemini.google.com. OpenAI says ChatGPT tags many of those links with a UTM source of chatgpt.com, which makes them easier to spot. GA4 lets you build a custom channel group with a "matches regex" rule so these domains roll up into one "AI assistants" channel, and Google's own help page uses that example.

Two caveats apply. Some AI visits arrive without any referrer and get counted as direct traffic. And clicks from Google AI Overviews or AI Mode are counted as Google organic traffic, so GA4 alone cannot tell you which Google clicks came from an AI answer.

Source	What it shows	What it misses
Search Console Generative AI report	Impressions in AI Overviews, AI Mode, Discover	Clicks, and the split between Overviews and AI Mode
Bing Webmaster Tools AI Performance	Citations in Copilot and Bing AI summaries, grounding queries	Clicks and visits
GA4 referrals	Visits and conversions from AI assistants	Visits with no referrer, AI clicks inside Google

Run your own prompt tracking

None of the dashboards above tell you what the AI actually said about you. For that you need prompt tracking: a fixed list of questions you run on a schedule and log.

Build the prompt list. Aim for 20 to 40 questions written the way a real customer would ask them. Mix these types:

Category plus location: "best med spa for laser hair removal in Scottsdale"
Cost and logistics: "how much do clear aligners cost in Austin", "dentist open Saturdays near downtown Tampa"
Comparison: "Invisalign vs braces cost for adults", "[your brand] vs [competitor]"
Brand checks: "is [your practice] any good", "[your brand] reviews"
Buying intent for products: "best electrolyte powder without sugar"

Pull wording from your intake forms, call notes, site search and the questions your front desk hears every week.

Run it consistently. Use the same engines, the same account state (logged out or a fresh session) and the same location each time, and note the date. AI answers are not fixed. The same prompt can produce a different answer an hour later, so run each prompt two or three times per check and treat the results as a sample, not a ranking.

Keep the cadence realistic. Monthly is enough for most small businesses. Run an extra check after a big site change, a new location or a new product launch.

What to record for each answer

A spreadsheet with one row per prompt, per engine, per run works fine. These are the columns that matter:

Field	What to note
Mentioned	Did your business or brand appear at all?
Position	First recommendation, one of several, or a passing mention
Cited link	Which URL was linked, if any: your site, Google Business Profile, a directory
Accuracy	Hours, prices, services, location, providers: anything wrong?
Framing	Positive, neutral or negative, and any outdated complaint repeated
Competitors	Which other businesses were named
Sources used	Third-party pages the engine cited: review sites, directories, Reddit, local news

From that log, a few numbers are worth tracking month to month:

Mention rate: the share of prompts where you appear.
Citation rate: the share where your own site is linked.
Share of voice: your mentions compared with your top three competitors on the same prompts.
Error count: factual mistakes about your business.
AI referral visits and conversions: from your GA4 channel group.

The error count is often the most useful for a local practice. An assistant that tells people you are closed on Saturdays, or lists a service you stopped offering, costs bookings no matter how often it mentions you.

Where paid tracking tools fit

Running 30 prompts across five engines by hand takes a few hours a month. Paid tools automate it: they run your prompt list on a schedule, log mentions and citations, and chart share of voice against competitors. Options in this category include Semrush's AI Visibility Toolkit, Ahrefs Brand Radar, Otterly.AI, Peec AI and Profound. Pricing, engine coverage and prompt limits change often, so compare the current plans directly.

Keep two things in mind. These tools are sampling the same changeable answers you would see manually, so their numbers are estimates. And Google's own guidance warns against third-party tools that claim access to internal Google metrics, because none of them have it. A manual spreadsheet for the first two or three months is a good way to learn which prompts matter before you pay for automation.

Turning the data into action

Measurement only helps if it changes what you do. Common patterns:

Wrong facts: update your Google Business Profile, your website and your main directory listings so they agree. Assistants often repeat whatever the most-cited source says.
Competitors cited from third-party pages: look at which directories, review sites or articles the engines lean on, and make sure your listing on those sites is complete and current.
Impressions in Search Console but few clicks: check whether the pages being shown answer the question clearly enough that people want more.
Not appearing anywhere: check that robots.txt is not blocking OAI-SearchBot, PerplexityBot or Bingbot, and that your site has not been opted out of generative AI features with the toggle Google added to Search Console in 2026. Google says opted-out sites get no traffic or impressions from those features.

The how-to side of this, from page structure to crawler access, is covered in our GEO best practices checklist.

A monthly routine

Export the Search Console Generative AI report and note the top pages.
Check Bing AI Performance for citations and grounding queries.
Pull AI referral sessions and conversions from your GA4 channel group.
Run your prompt list on each engine and fill in the log.
Fix any factual errors at the source.
Pick one or two pages or listings to improve before next month.

Where NutriCove fits

NutriCove builds websites and web apps with lead capture, CRM and booking wired in. Our multi-page builds include Google Business Profile and analytics setup, on-page SEO and schema markup, which gives you the baseline data described above. If you are planning a new site or a rebuild, see how we build websites.

Sources

This article is general information, not legal or medical advice. Talk to a qualified attorney or licensed professional about your situation.