You can find out whether AI systems cite your business in ten minutes, for free, with no paid tool. Two things most owners treat as one come apart here: being named (your business appears in the words of the answer) and being cited (your page is attached as a clickable source). The audit is three moves. First, ask ChatGPT, Copilot, Perplexity, and Google's AI Overview the exact question a buyer types before calling you, and record for each whether you were named, cited, or neither. Second, open two free first-party dashboards, Bing Webmaster Tools' AI Performance report and Google Search Console's generative AI report, to see what those systems already did with your existing pages. The dashboards register only a link, so an answer that names you without linking you is invisible to both; the manual test is the only way to catch that. Zero data is itself an answer. Record the grid the same way every month and you can watch it move.
Most owners have never run this check, so they are guessing. Guessing is expensive here, because the answer moves: an AI system that ignored your page in March may cite it in July, and the reverse. The flagship in this series proved with our own tracking that ranking and citation are separate outcomes on the same question, measured the same day. This piece is the hands-on version: the exact ten-minute routine to find out where your business stands, using nothing but a browser and two free dashboards you may already own.
Named and cited are two different things
Named and cited are two outcomes, not one, and the whole audit turns on telling them apart. Named means your business name appears in the words of the AI answer. Cited means your page is attached as a clickable source: a numbered footnote, an inline citation, or an entry in a Sources panel. They come apart in both directions. An answer can name you in a sentence and link you nowhere, and it can list your page as a source without ever saying your name in the prose.
That distinction is not academic, because the two free dashboards you will open in minute five register only a link. When an AI answer names your business in the sentence but attaches someone else's page as the source, or no source at all, neither dashboard records a thing about it. Our flagship test caught exactly this: asked a home-services hiring question, Perplexity named three competitors in the body of its answer and did not cite us, even though our site sat in its pool of ten retrieved sources. A dashboard that counts links would have told us nothing about that answer. The manual test is the only way to see mention-without-link.
Minutes one to five: ask four engines your customer's question
Type one question, the exact one a buyer types before calling you, into four engines and watch where each one points. Write it the way a real customer would, not the way a marketer would: the plain version ("who should I hire for [your service] in [your city]") and the longer, messier version a real buyer pastes in. Run it in all four, one after another, and for each one note two things: were you named, and were you cited.
| Engine | Where to run it (free) | Where it shows sources | What to record |
|---|---|---|---|
| ChatGPT | chatgpt.com, works logged out | Inline citations you hover to preview and click to open; if none appear, a Sources panel beneath the response | Confirm the Search tool is on so it grounds the answer rather than answering from memory, then note named vs cited |
| Microsoft Copilot | copilot.microsoft.com | Numbered citations and a sources list under the answer | Whether your domain is in that list; this is the exact surface Bing's AI Performance report tracks |
| Perplexity | perplexity.ai, free tier | Clickable citations in every answer, with sources numbered at the top | That you can sit in the retrieved source list and still not be the source it built the sentence from |
| Google AI Overview | google.com | The expanded AI Overview shows linked source cards | Whether your site appears as a linked card, and separately whether your name is in the summary text |
Two engine quirks are worth a note before you trust the result. ChatGPT's search responses only include inline citations when the Search tool is active; if you do not see any, turn Search on and ask again. Perplexity puts clickable citations on every answer, which makes it the clearest illustration of the named-versus-cited gap: read its numbered sources and check whether your domain is actually one of them, because being retrieved is not the same as being quoted.
Minutes five to ten: open the two free dashboards
Two search platforms now publish, for free, exactly which of your pages their AI systems used as evidence. Together they are the first-party half of the audit, and they cover what the manual test cannot: what these systems already did with your site across many real queries, not just the one you typed.
Bing went first. Bing Webmaster Tools added an AI Performance report on February 10, 2026, a public preview separate from the classic Search Performance report, and it requires a free Bing Webmaster Tools account with your site verified. It reports Total Citations, cited pages (and an average), grounding queries, page-level citation activity, and which grounding query mapped to which page. One limit matters above all the others: it covers Microsoft Copilot, Bing's AI-generated summaries, and select partner integrations. It does not cover ChatGPT. Anyone who tells you a Bing dashboard measures your ChatGPT presence is wrong. The grounding queries are generalized, sampled theme phrases, not the real prompts users typed, so read them as themes rather than transcripts. Data refreshes daily, is sampled (low-volume activity may not surface), and exports to CSV or Excel across 7-day, 30-day, 3-month, or custom ranges. Microsoft's June 16, 2026 update added Intents, Topics, Citation Share, and a Compare view; treat Citation Share with care, because Microsoft itself calls it observational and not a market-share number.
Google added the second dashboard on June 3, 2026: a dedicated generative AI performance report in Search Console. It is still rolling out to a subset of properties, so you may open Search Console and not see it yet; Google says so plainly. It covers AI Overviews and AI Mode on Google Search only, not ChatGPT, Copilot, or Perplexity. And here is the trap hiding in the name: it measures impressions, not citations.
The report breaks down by Pages, Countries, Devices, and Dates only. Unlike Bing's grounding queries, there is no query or keyword dimension, so it will not tell you which questions surfaced your pages. It has an Export button and caps the table near 1,000 rows.
If either dashboard is empty, that is not a bug; it is the finding. Empty means those AI surfaces are not currently using your pages as evidence. Bing lists the likely reasons: your content was less relevant, lacked clarity or depth, was outdated, or was outperformed by other sources. For Google, empty can also mean the report has not reached your property yet, you do not have enough impressions, or you excluded your site from Google's generative AI features, which removes you from both the report and the answers themselves.
Record it the same way every month
A check you run once is trivia; a check you run the same way every month is a baseline. The recording method is deliberately small: one row per engine, three columns. Was your business named in the answer, was your page cited as a source, and who got named instead.
| Engine | Named? | Cited? | Who got named instead |
|---|---|---|---|
| ChatGPT | |||
| Microsoft Copilot | |||
| Perplexity | |||
| Google AI Overview |
Fill it on the same day, then repeat the identical run next month. The "who got named instead" column is the one that pays off: across three or four months it shows which competitors the machines trust for your buyer's question, and whether that set is shifting toward you or away. This is the owner-scale version of what we do continuously. We run this cross-engine test on our own business three times a week, measured the same day, as part of the 60,000+ data points we monitor daily.
Third-party trackers exist if you outgrow the manual grid. We use Semrush to track the same buyer questions across Google, Google's AI answers, and ChatGPT on a running basis, and other AI-visibility tools are entering the market. Start with the free grid and the two dashboards anyway; you learn more running the test by hand than from a tool that hides the answer behind a single score.
A "no" is a fixable finding, not a verdict
A row of "no" does not mean AI has blacklisted your business; it means your pages did not give the machines anything to lift. If the manual test named your competitors and your most important service page has no single sentence that states your business, your service, your city, and one checkable fact together, those two findings are the same finding. The engines had nothing standalone to quote, so they quoted someone who did.
That is an evidence problem, and evidence problems are fixable. The fix is rewriting pages into passages that stand on their own as evidence, which is its own discipline and the subject of a separate piece in this series on writing evidence instead of marketing copy. The audit tells you where the gap is; that work closes it.
The point of the ten-minute check is not the score. It is the direction. Run it, record it, watch the grid move month over month, and fix the pages where the "no" columns cluster. If you would rather we run it for you and hand back the grid plus the fix list, ask us; either way, look at both scoreboards before your competitors do. This is what our SEO practice now tracks on every client: named, cited, and the gap between them.