Reader guide
Turn this article into action
Plain-English point
Google Search Console now has a dedicated Generative AI report. Here is what it shows, what it cannot, and a five-step workflow that combines it with prompt-level tracking to find and fix the queries where AI answers skip your brand.
Best use
Use this article when you are working on geo basics and need a concrete GEO next step.
TL;DR: Google now gives you a dedicated Generative AI report in Search Console (launched June 2026) that shows impressions in AI Overviews and AI Mode. It tells you how often your pages appear, but not what the AI said or who else was cited. For that, pair it with prompt-level tracking of the queries where you already have impressions.
The measurement problem, and what changed
For a long time, AI Overviews were a blind spot. Google confirmed in 2025 that AI Mode data was folded into the normal Web totals in Search Console, so you could see traffic but could not separate it from classic blue-link results.
In June 2026, Google introduced dedicated Search Generative AI performance reports covering AI Overviews, AI Mode and generative features in Discover. You can break the data down by page, country and date.
Rollout was staged: Google first described it as reaching a subset of sites, and third-party coverage reports wider availability since. If you do not see the report on your property yet, check again later and confirm you have a verified property.
What Search Console tells you, and what it cannot
| You can answer | You cannot answer |
|---|---|
| Which of my pages appear in AI features | What text the AI actually wrote |
| How impressions trend over time | Whether my brand was named in the answer |
| Which countries and dates drive them | Which competitors were cited alongside me |
| Which queries are growing | Whether the sentiment was positive |
That second column is the one that decides whether a visit turns into a customer. Appearing as a small source link under an answer that recommends a competitor is a different outcome from being the recommended brand.
A practical workflow
Step 1: Pull the queries where you have impressions but no clicks
In the Search Console Performance report, filter to queries with meaningful impressions and a low click-through rate. Long, question-shaped queries are the likeliest to trigger AI Overviews.
For MettleGEO's own site, this surfaced queries like "ai overviews mentions tool" (about 150 impressions in 30 days, average position 28, zero clicks). That is a query with real demand where we do not yet have a page that deserves the answer.
Step 2: Check what the AI answer actually says
Take the top 20 to 30 of those queries and run them as prompts through an AI search tracker. For each, record whether the answer exists, whether you are named, and which domains are cited.
This is the step Search Console cannot do for you, and it is where the cross-reference becomes valuable.
Step 3: Sort the queries into four buckets
AvoidAI answer exists, you are not cited
PreferContent gap: build the page type the answer is quoting
AvoidAI answer exists, you are cited but not named
PreferEntity gap: strengthen brand clarity, About page, schema
AvoidNo AI answer, you rank poorly
PreferClassic SEO gap: standard ranking work first
AvoidNo AI answer, you rank well
PreferProtect it: keep the page fresh, watch for an AI answer appearing
Step 4: Fix the page, not the prompt
AI Overviews pull from pages that answer the question cleanly. For each content gap:
- Put a two to three sentence direct answer immediately under a question-style heading.
- Add a comparison table or numbered steps where the query implies one.
- Include specific numbers, dates and named sources.
- Add FAQ or HowTo structured data when the page genuinely contains them. See our guide to schema markup for AI.
Step 5: Measure again in 4 to 6 weeks
Re-run the same prompts and compare. Look for movement in three numbers: share of queries where an AI answer appears, share where you are cited, and share where you are named.
Do not over-read one week
AI answers change between runs. A drop in a single week is noise unless it persists across several runs of the same prompt.
Why Overviews citations do not track your rank
It is tempting to assume that ranking in the top 10 gets you into the Overview. The evidence is mixed and shifting. Ahrefs reported that only about 12% of URLs cited by assistants such as ChatGPT, Gemini and Copilot ranked in Google's top 10 for the prompt, and published numbers for how closely Overviews follow the top 10 vary between studies. The safe reading: ranking helps, but a page that answers the question in the cleanest, most quotable way can be cited without ranking first.
How Mettle connects the two
When you connect Google Search Console, Mettle cross-references your top queries with its own AI answer data and flags queries that have impressions but no AI citation. Those are the highest-value content and schema fixes. The list is available in the dashboard and through the Mettle MCP connector, so you can ask your assistant for the top missed queries and draft the page in the same conversation.
One honest limit: Mettle tracks ChatGPT-style (OpenAI), Gemini and Perplexity answers through prompt runs, and uses Search Console for Google's own AI features. It does not scrape Google's AI Overviews directly, so Search Console remains your source of truth for Overview impressions.
Questions this article answers
Does Search Console report AI Overviews separately?
Yes. Google introduced Search Generative AI performance reports in June 2026, covering AI Overviews, AI Mode and generative features in Discover, with breakdowns by page, country and date. Rollout was staged, so check your property.
Can Search Console tell me what an AI Overview said about my brand?
No. It reports impressions in AI features, not the answer text, whether your brand was named, or which competitors were cited. Prompt-level tracking fills that gap.
Which queries should I check first?
Long, question-shaped queries with high impressions and low click-through. These are the likeliest to trigger AI answers and the most valuable content gaps to close.