Google AI Overviews tracking is the measurement layer that reports how often Google's AI Overview cites your site, and which sources it cites instead. An Overview names a handful of sources and resolves the question above the links; this tracks whether you are one of them, across your buyers' real questions, on a schedule.
An engine named after a surface it did not measure is a fabricated result with extra steps, so here is the precise scope before anything else. For Google AI Overviews, we measure the AI Overview block on a live Google results page, together with the sources it cites.
| What it means here | |
|---|---|
| Direct evidence | The AI Overview text and its cited-source list are read from the surface itself. That is direct evidence for this engine. |
| Not claimed | Google AI Mode is a second surface over the same Google index. A reading from one is not reported as the other, because a shared index does not make them the same answer. |
| How a run works | We request the Google results page through a SERP provider and read the ai_overview block: its text, and the list of sources it links. Nothing is inferred — if Google returns no AI Overview for a query, the run records that it did not, rather than guessing. |
Google AI Overviews tracking is also known as AI Overview monitoring or SGE tracking. The same direct-versus-proxy distinction is applied to every engine we support, and it is written into the published methodology rather than left to a sales conversation.
Rates, not screenshots — each with a 95% confidence interval and the number of answers behind it, so you can tell a real change from a noisy week.
| Metric | The question it answers |
|---|---|
| Mention rate | Of the answers to your category's questions, how many name you at all |
| Share of voice | Of every brand named across those answers, what fraction is you |
| Recommendation rate | How often you are actively endorsed rather than listed in passing |
| Cited sources | Which domains Google AI Overviews leaned on — yours, and everyone else's |
| Accuracy flags | Anything Google AI Overviews said about you that is factually wrong |
One reading is not a trend. Runs repeat on a schedule so the comparison is against your own history, not against a screenshot someone took once.
Yes, and it is worth tracking separately. Google does not return an AI Overview for every query, and whether one appears can change week to week. A run records the presence or absence of the block as its own fact, so a drop in your mention rate can be told apart from the Overview simply not firing.
No. AI Mode is a different surface built over the same index, and it can answer the same question differently. Because they are different surfaces, a reading from one is not reported as if it came from the other.
Because the source list is where the competitive picture is. The text tells you what Google said; the sources tell you which domains it trusted to say it, which is the list you are trying to join.
The editorial companion to this page — how Google AI Overviews chooses what to say, and what earns a mention.
Google AI Overviews is one surface. The full tracking layer runs your suite across all of them and compares.
The sampling, the calibrated judge, the statistics — and the six-point bar to hold any tool to.
Run your category's real buyer questions and get a rate with an interval — not a screenshot.