Spoke
First-party AI visibility reporting
Google and Microsoft now report AI visibility directly, and OpenAI tags its referrals. That beats a third-party citation tracker — and each one measures something narrower than its name suggests.
What the Google report contains and what it leaves out comes from Google's own announcement, retrieved 15 November 2026. The Microsoft and OpenAI items are documented by those platforms, described at the level we could verify.
Not long ago the honest answer was that AI visibility had no first-party reporting at all, and you had to approximate it with a panel of prompts.
That's no longer the whole picture. The update is genuinely good news with a sharp limit attached.
Can you now see how often you appear in AI Overviews?
Yes, as impressions. Google's Search Console generative AI reports show how often your URLs turned up in AI features in Search and Discover, broken out by page, country, device and date. That data also feeds the overall performance report.
What you don't get is clicks for those features, or the questions behind them. So you can see that you appeared and roughly how often. You can't see what it produced or what anybody asked.
What each one actually covers
| Source | Covers | Reports | Doesn't report |
|---|---|---|---|
| Search Console generative AI reports | AI Overviews, AI Mode, generative features in Discover | Impressions by page, country, device, date | Clicks or queries for those features |
| Bing Webmaster Tools AI Performance | Microsoft's own AI experiences | Citations, cited pages, grounding queries | ChatGPT, Perplexity, or Google's features |
| ChatGPT referral tagging | Visits arriving from ChatGPT | Referral traffic you can segment | Impressions, or citations with no click |
Read the last column before the middle one. Each of these gets described as measuring "AI visibility." Each measures one vendor's slice of it.
The Google report, precisely
It shows impressions of your URLs in AI features across Search and Discover. Breakdowns by page, country and date, plus device for Search. It's folded into the overall performance report rather than living in its own universe.
Two cautions matter more than the feature itself.
An impression isn't a visit, a citation, or a sale. It's an appearance. The zero-click gap this section keeps describing is exactly the space between that number and anything commercial. The report doesn't close that gap. It measures it.
No query data. You can see which pages appeared. You can't see what people asked, which is the thing you'd most want for planning what to write next.
There's a related control worth knowing about: a Search Console setting for whether your site is included in Search's generative AI features at all. It's separate from Google-Extended, which governs training and grounding elsewhere and which Google documents as not affecting inclusion in Search. Opting out removes you from those features, and the impressions go with you. That's a clean trade — just make it deliberately, rather than by clicking something that looks like a privacy setting.
Microsoft's version
Bing Webmaster Tools reports how your content shows up in Microsoft's AI experiences: citations, which pages got cited, and the queries behind the retrieval, with later additions covering intent, topics and citation share.
The qualification Microsoft itself makes is the useful part. Citation activity isn't a measure of ranking, authority or how important your page is. It's a count of appearances. Treating a rising count as proof you're getting better is the same mistake as treating impressions that way.
ChatGPT referrals
OpenAI tags visits arriving from ChatGPT with an identifiable source, so they show up as their own segment instead of disappearing into direct traffic.
This is the most useful of the three and the narrowest. It catches people who clicked. It tells you nothing about the much larger group who read an answer citing you and didn't — which, as this whole section argues, is most of the value and most of the problem.
What changes and what doesn't
Changes: you no longer have to rely on third-party citation trackers of unknown method for Google and Microsoft. First-party data beats an inferred estimate, and it's free.
Doesn't change: the attribution gap. Impressions and citations still won't tell you what business came out of it. Keep running the prompt panel, because it covers the engines that report nothing — and because it records what was actually said about you, which none of these reports do.
Worth repeating: three partial views don't add up to a total. Anybody presenting you a single "AI visibility score" built out of these is adding incompatible units from vendors measuring different things in different ecosystems.
More in Measurement
Prompt panel tracking
Asking an assistant about yourself once tells you nothing. A fixed set of prompts, re-run on a schedule and recorded, tells you a direction — which is the most that is available.
The attribution gap
Being cited without being clicked is a real outcome that last-click reporting cannot see. It usually resurfaces later as direct or branded traffic, credited to the wrong thing.
AI share of voice
How often you are named across a fixed set of questions, relative to the people you compete with. It is noisy, it is the closest thing available, and it breaks the moment anyone reports it as a single number.
AI referrers
The visitors who do click from an AI answer are identifiable in ordinary analytics. It is the easiest measurement win available here, and it counts the smallest part of the channel.
Self-reported attribution
One required field on every intake. It is biased, imprecise, and the only place a human tells you directly that a machine recommended you.
KPIs that survive
Sessions as a headline metric tells a story that gets less true every quarter. These are the measures that keep meaning what they meant when the traffic mix changes underneath them.