How to Track Google AI Overviews

Tracking Google AI Overviews comes down to answering three questions on a schedule: does an AI Overview appear for the queries your buyers use, does it cite you, and who else does it cite? Google now answers a slice of the first two itself — Search Console added a generative-AI performance report in June 2026 — but that report shows impressions only, for your site only. The third question, the one that actually changes decisions, Google will never answer for you. This guide walks through what each layer of tracking shows, what it hides, and how to build a monitoring habit that catches the moment an answer starts naming a competitor instead of you. (If you first need the plain-English explanation of what AI Overviews are, start with the definition post.)

TL;DR

• Search Console now has a generative-AI performance report: impressions your URLs earned inside AI Overviews and AI Mode, by page, country, device, and date. No clicks, no CTR, no query breakdown — and the two AI surfaces are reported together, not separately.
• GA4 can't isolate AI Overview clicks. They arrive from google.com like any other search click, so they land in Organic Search unlabeled.
• No Google tool shows competitors. Seeing when an Overview cites someone else — and which of their pages it pulled — requires manual checks or third-party tools.
• The workable stack is three layers: Search Console for your own baseline, a rank tracker that flags AI Overviews per query, and an AI-visibility tool for the competitive, cross-engine picture.
• Track the trend, not a snapshot. AI Overviews change as Google's models and the live web change; a monthly cadence is the floor.

What does Search Console show about AI Overviews?

Two things — one blended, one broken out.

The main performance report has always counted AI Overview appearances and clicks inside your normal totals. If your page is cited in an Overview and someone clicks it, that's an impression and a click like any other — you just can't tell which rows came from an Overview and which from a blue link.

The generative-AI performance report, rolled out from June 3, 2026, is the first view that separates AI from classic search. It shows how often links to your site appeared inside Google's generative-AI features, with the pages involved and breakdowns by country, device, and date. That's a real unlock: for the first time you can see what share of your Google visibility is AI-feature visibility.

Three limitations to know before you open it, straight from Google's documentation:

1. Impressions only. No clicks, no CTR, no position. You learn that you appeared, not whether anyone acted on it.
2. No query dimension. You can see which pages surfaced in AI features but not for which searches.
3. AI Overviews and AI Mode are combined. The report covers both surfaces as one number, so you can't tell a summary-on-a-results-page impression from a conversational AI Mode one.

It's also on a gradual rollout — some properties don't have the report yet, and low-volume sites may not see data. If you're using it to diagnose a traffic drop rather than to build a tracking habit, we've written that step-by-step diagnostic separately — this post won't repeat it.

What does Google's own reporting still hide?

Two blind spots, and the second is the expensive one.

First, the click gap. Between Search Console's impressions-only AI report and GA4's attribution, you cannot cleanly measure what an AI Overview appearance is worth in visits. GA4's new AI Assistant channel doesn't help here either — it captures referrals from ChatGPT, Gemini, Copilot and others, but a click from an AI Overview arrives from google.com like any other search click and files under Organic Search, indistinguishable from the blue link it replaced.

Second, the competitive gap. Search Console only ever reports your property. It cannot tell you that an AI Overview for your money query exists at all if you're not in it, that it named three competitors, or which of their pages it cited as sources. That's the information that redirects content and PR effort — and no Google report will ever contain it, because Google has no reason to show you someone else's data. Everything beyond this point in the stack exists to close that gap.

How do you track AI Overviews manually?

Start free, today, with a spreadsheet — the method takes about an hour a month and tells you whether the problem is even worth tooling.

1. List 10–20 questions your buyers actually ask. Not keywords — questions: "best [category] for [situation]", "is [your product] worth it", "[you] vs [competitor]". These long, question-shaped queries are exactly where AI Overviews trigger most often.
2. Search each one in Google and record four columns: did an AI Overview appear, are you named or cited in it, who else is cited, and what the answer actually says about your category.
3. Repeat monthly, same queries. The value is in the deltas — an Overview appearing where there wasn't one, a competitor entering a citation list, the answer's framing shifting.

Two honest caveats. AI Overviews vary by location, device, and moment — your check is one sample, not the truth. And a manual check doesn't scale past a few dozen queries. But it costs nothing, and it answers the first strategic question — are AI Overviews an opportunity or a threat for us? — faster than any procurement cycle.

What tools track AI Overviews at scale?

Think in three layers, cheapest to richest. They stack rather than replace each other.

Layer 1: Search Console. Free, first-party, your baseline for AI-feature impression share and the pages earning it. Limits as above: your site only, impressions only.

Layer 2: a rank tracker with AI Overview detection. Most major rank trackers — Semrush and seoClarity among them — now flag, per tracked keyword, whether the SERP includes an AI Overview and whether your domain is cited in it. This scales the manual check to hundreds of queries and adds history, so you can see Overviews arriving on your keywords over time. What it tracks is the SERP feature, though — Google only.

Layer 3: an AI-visibility platform. This layer answers the competitive and cross-engine questions the first two can't: who else is cited for your buying questions, which source pages the answers pull from, how you're described — and how the same questions resolve in ChatGPT, Perplexity, and Gemini, where a growing share of the same research happens with no SERP at all. We've compared what these tools measure and how to pick one separately; ClappX AnswerX is ours, and it runs your category's buying questions across Google's AI surfaces and the major assistants continuously, flagging the answers where competitors appear and you don't.

The right stopping point depends on stakes. If AI Overviews rarely trigger on your queries, layer 1 plus a quarterly manual check is enough. If they sit on your money queries and cite your competitors, the third layer is where the actionable signal lives.

How often should you check — and what counts as a change worth acting on?

Monthly is the floor for most brands; competitive categories justify weekly or continuous monitoring. AI Overviews are rebuilt from a live index by models Google keeps updating, so answers drift — which is exactly why one-off audits mislead. Track four things over time:

• Trigger rate: what share of your tracked queries show an AI Overview at all.
• Citation rate: on queries with an Overview, how often you're cited or named.
• Competitor share: who else keeps appearing, and whether new names are entering.
• AI impression share: in Search Console, generative-AI impressions as a share of total impressions, and whether clicks on the same pages are holding.

Act on sustained shifts, not single readings: a competitor cited three months running is a signal; one odd answer on one afternoon is noise. And when the citation rate on queries you care about is falling, the fix is rarely inside the tracking stack — it's answer-shaped content and third-party evidence, the work the tracking exists to prioritize.