Skip to content
Built for AI agentsConnect yours
Google AI Overviews logo

Google AI Overviews

AI Overview tracker: get recommended in Google AI Overviews

Google AI Overviews answer the question before anyone reaches a result. See which Overviews recommend a competitor instead of you, the pages Google cited to get there, and whether a change moved them, with the queries that showed no Overview kept apart.

Start 7-day free trialAll platforms

Card required, nothing charged for 7 days

  • Live SERP
  • Six days a week
  • Asked from your market

Live SERP

No Overview on a query is a finding, not a loss

We read Google's live results page six days a week, from the market your workspace measures. Google shows an Overview for some queries and not others, so when none appears the answer is marked “no AI answer” and left out of every rate. A query with no Overview has nothing to win yet; it is not one you are losing.

The live results page.
Read as Google serves it for each query, so the Overview is the one a searcher would see.
From your market.
Google localises its Overviews, so we search from the country your workspace measures, and a US result never mixes with a UK one.
Absences kept out.
A “no AI answer” row sits outside every denominator, so Google's choice not to show an Overview never reads as your absence.

Every check, one of four

Four outcomes, not one score

Every query lands in one of four states on every check. Collapsing them into a single percentage hides the only thing that tells you what to do next.

  1. no AI answer

    No Overview shown

    Google answered with ordinary results. Nothing to win on this query yet, so it is left out of your rate rather than counted as a loss.

  2. not named

    Overview shown, you are absent

    An Overview appeared and named other brands. This is the real gap, and the number most tools quietly blend into an average.

  3. cited, not named

    Cited, not named

    Your page is in the source panel but a competitor is named in the prose. You have the retrieval and not the framing.

  4. named you

    Named

    Your brand appears in the answer text. We record where it came among the brands named, because first and last are not the same outcome.

Example dataGoogle AI Overview answers as the app opens them, under the verdict the Answers screen gives each.

Track AI Overviews

How to track AI Overviews: what to measure on every check

An Overview answers the question above the results, so a search can end without a click. Ahrefs measured a 34.5% drop in clicks to the top organic result when one is present, SparkToro found that only 374 of every 1,000 US Google searches end with a click to the open web, and Seer Interactive found organic click-through falling year on year on the queries where Overviews show. That makes them worth tracking, and almost every habit carried over from rank tracking misleads here.

Recorded on every check, per query

  • Whether an Overview appearedThe base rate. If Overviews show on three of your twenty queries, your surface is three queries, and every rate is taken against it.
  • Whether your brand was namedThe headline, as a share of the checks where an Overview appeared, never as a share of all checks.
  • Whether your page was citedKept apart from named. The gap between the two is the most actionable signal an Overview gives you.
  • Which competitors were namedNamed in two Overviews of twenty means one thing if nobody else is named, and another if one rival is named in eighteen.

How often, and for how long

  • Several checks a weekWeekly is too slow to tell a real move from ordinary variance, and hourly buys volume that tells you nothing new. We check each query six days a week.
  • Two weeks before a directionThe first day tells you where you stand. The first fortnight tells you which way it is going, and only that is worth acting on.
  • One market, fixed wordingAn Overview served in London can name different brands to one served in New York, and a reworded query starts a new series.
  • Ten to thirty real queriesPhrased the way your buyers type them, with your competitors tracked on the same set.

Common mistakes when tracking AI Overviews

Each one moves the number without anything changing in the Overview.

Common mistakes when tracking Google AI Overviews, what each does to the number, and what to do instead
Checking once and believing itDoes:Reports one draw from something that varies as though it were a fact.Instead:A rate across repeated checks, with the number of checks beside it.
Counting queries with no Overview as lossesDoes:Drags your rate down with queries that were never winnable, and sends you after a problem that is not there.Instead:Record no Overview as its own outcome and leave it out of the rate.
Merging citations and mentions into one scoreDoes:Hides cited, not named, whose fix is the opposite of the fix for being absent.Instead:Report named and cited as separate states.
Ignoring competitorsDoes:Leaves your own number with nothing to say whether two in twenty is good or bad.Instead:Track the rivals you lose to, on the same queries and the same days.
Not checking snippet eligibility firstDoes:A stray max-snippet:0, or data-nosnippet around the paragraph that answers the question, keeps a page out of every Overview while its blue-link rank stays put. No tracker shows it.Instead:Read your robots meta tags and snippet controls before you read any number.
Rewording a tracked queryDoes:Starts a new series that only looks like the old one.Instead:Fix the wording. If it has to change, treat it as a new query.

Google documents the mechanism itself: Overviews are assembled with retrieval-augmented generation and query fan-out, and shown only when its systems judge them additive to classic Search, so on many queries they do not appear at all.

Overview text and citations

What the Overview said above the results, and what it cited

When an Overview does appear, we keep its text verbatim, the brands in the order it named them, and every page in its citation panel. The panel and the prose are separate questions: Google can link your page in the panel and still recommend a competitor in the text people read.

The prose.
What a searcher reads before any link, word for word.
The citation panel.
Every page Google linked under the Overview, in order, with your own pages marked.
The order it named brands.
First mention and last mention are not equivalent, so the position of every brand is kept.

Repeated, not once

One check of an AI Overview is a coin flip

AI Overviews are generated fresh on each search rather than read from an index. Two checks minutes apart can name different brands, cite different pages, or produce no Overview at all. A tool that checks once and reports a green tick is reporting a coin flip. The same rule applies across every assistant we track.

Six days a week, per query.
The same query on a fixed schedule, from the market you chose, so the rate you see is a rate and not a snapshot.
Fractions below five.
Under five checks we show 2 of 4, not 50%. One run is a draw, not a rate.
Absence is not a loss.
A day with no Overview leaves the denominator rather than counting against you.

Per-engine visibility

AI Overview tracking, counted only where an Overview appears

The Overview's count is the number of answers where an Overview actually appeared, which is why it is often smaller than the other engines' for the same prompts. It sits beside AI Mode as a separate engine, and is never averaged with Gemini, ChatGPT, Claude or Perplexity.

Why AI Overviews and AI Mode are tracked as two engines

Both are Google. They retrieve differently and often disagree.

AI Overviews sit above ordinary results on a normal search. AI Mode is a separate conversational surface with its own retrieval, and has its own page for Google AI Mode tracking. Google says to expect a difference: the two may use different models and techniques, so the responses and links they show will vary. We have measured them naming different brands for the same question on the same day, so merging them into one Google score would average away the disagreement that matters. They are collected and reported separately, in line with the rest of our methodology.

How a single blended Google figure and RankHound each report the same Google checks
AI Overviews and AI ModeBlended:Averaged into one Google figure.Kept apart:Two engines side by side, each with its own rate, trend and sample size.
A query with no OverviewBlended:Counted as a miss.Kept apart:Recorded as no AI answer, and left out of the rate.
Your page cited, a rival namedBlended:Lost inside the average.Kept apart:Reported as its own state: cited, not named.
Where the answer came fromBlended:Gone once the figure is computed.Kept apart:Collected from the live result page, with every cited page stored.

Before you trust an AI Overview tracker

If something here is unclear, ask us before you sign up rather than after.

Start 7-day free trial

Card required, nothing charged for 7 days

What is an AI Overview tracker?
Software that repeatedly checks Google AI Overviews for a set of queries and records whether an Overview appeared at all, which brands it named, and which pages it cited. It differs from a rank tracker because an AI Overview is not a ranked list: your brand is either named in the prose or it is not, and the citation panel is a separate question again. RankHound does this as part of an AI visibility platform built to get you recommended, so it also shows the pages to work on and whether that work changed the Overview.
How do I track AI Overviews?
Choose the queries your buyers type, fix their wording, and check each one on a schedule from one market. On every check, record four things: whether an Overview appeared at all, whether your brand was named in it, whether your page was cited in its panel, and which competitors were named. Count your rate only over the checks where an Overview appeared, and keep your competitors beside you so the number means something. RankHound runs those checks six days a week from the market your workspace measures and keeps every Overview in full.
How do I get recommended in Google AI Overviews?
Start with the Overviews that named a competitor, and the pages in their citation panels: those are the pages Google drew on. The Sources screen flags the sites that named a rival every time and never you, which are the places to get onto. If your own page is cited and a rival is named, the work is in how that page frames you. The same queries are checked six days a week, so the rate shows whether a change moved the Overview.
Can RankHound tell me when a Google AI Overview names my brand?
Yes. Your queries are checked six days a week from the market your workspace measures. We record whether an Overview appeared, which brands it named and which pages it cited. Collected from the live SERP.
What happens when Google shows no AI Overview?
The response is marked “no AI answer” and left out of every rate, because Google declining to show an Overview is not your absence. Overviews do not appear for every query, and whether one appears at all is the first thing worth measuring. A query with no Overview is not a query you are losing, so the absence is kept as its own outcome and never counted against you.
Why does the same query give a different Overview each time?
Overviews are generated, so two checks can differ in wording, in which brands are named and in which pages are cited, and on some days Google shows no Overview at all. The rate counts only the days one appeared, and shows the count beside it, as a fraction until there are five.
How often should an AI Overview be checked?
Several times a week for a set of queries you care about. Weekly is too slow to separate a real move from ordinary variance, and hourly buys volume that tells you nothing a steady series does not. We check each query six days a week. Give it about two weeks before reading a direction into the rate, and below five checks it stays a fraction, because one check is a draw, not a rate.
Can my site be cited in an AI Overview without my brand being named?
Yes. The citation panel and the prose are separate. Google can link your page as a source while naming a competitor in the sentence a reader actually reads. We report that as its own state because the fix differs: you already have the retrieval, what you lack is the framing the model repeats.
Does the country change the Overview?
It can. Google localises Overviews, so we search from the market your workspace measures. A workspace measures one country at a time, so an Overview seen in the UK is never blended with one seen in the US.
Do you combine AI Overviews and AI Mode into one Google score?
No. They are separate surfaces with separate retrieval, and they regularly name different brands for the same query on the same day. Merging them would average away the disagreement worth acting on, so each is reported as its own engine, and neither is averaged with the other four.

Related reading

7-day trial

Get recommended in the Overviews that leave you out

Add your brand, your competitors and your queries. Each is checked six days a week, the Overviews that named someone else are kept apart from the queries that had no Overview at all, and the weeks after show whether your changes moved them.

Start 7-day free trialHow to measure AI visibility

Card required, nothing charged for 7 days.

On every plan

  • Every answer stored in full, with its citations.
  • Every rate shown with the number of checks behind it.
  • Cited-but-not-named reported as its own state.