Where does AI Overview data appear in Search Console?

It is folded into your ordinary Performance data rather than broken out as its own report, with search-appearance filtering giving partial visibility. That means the impression you gained and the click you did not are in the same numbers as everything else, which is why the loss is easy to miss.

Which queries tend to get an AI answer?

Informational ones: definitions, how-to questions, comparisons, and anything with a short factual answer. Transactional and local queries are much less affected, because the searcher wants a provider rather than an explanation, and there is nothing for an answer to replace.

What does the pattern look like in the data?

Impressions holding or rising while clicks fall on informational queries, with position roughly unchanged. That combination is the signature. If clicks fell and impressions fell with them, something else happened and the answer feature is not your explanation.

Should I try to be cited in the AI answer?

Yes, and the work is the same work: direct answers near the top of the page, specific and sourced, on pages that are crawlable and fast. There is no separate optimization; being quotable is what earns the citation, and it also improves the page for humans.

What content should I stop writing?

Generic informational pieces whose whole value is explaining something an answer engine now explains in three sentences. That content had a traffic strategy behind it and the strategy has expired. The hours are better spent on pages that answer buying questions.

What should I do with existing informational content?

Reposition rather than delete. Add the specifics, the experience, and the position that a generated answer cannot supply, and connect it to the pages that convert. A generic explainer becomes useful again when it says something only your business can say.

Does this affect transactional content?

Much less. Queries where someone is choosing who to hire still send clicks, because an answer cannot make the decision for them. That asymmetry is the whole strategy: shift weight from explaining toward helping someone choose.

Is losing informational clicks actually bad for the business?

Usually less than the chart suggests. Traffic that arrived to read a definition rarely converted. The number to check is enquiries, not sessions, and a drop in one with the other flat is a reporting problem rather than a revenue one.

How do I separate this from a ranking drop?

Check whether position moved. An answer feature absorbing clicks leaves your position roughly where it was; a ranking drop moves it. Confusing the two sends teams rewriting pages that are still ranking exactly as well as they were.

Should I block AI features to protect my traffic?

You cannot block the answer feature without removing yourself from the search results it draws on, which trades a partial loss for a total one. For a business that wants customers, that is not a trade, and the option mainly exists for publishers with a different business model.

What is the practical response for a small business?

Concentrate on local and transactional queries, make every money page answer its question directly, and stop measuring success by informational sessions. That is a smaller change than the industry commentary suggests and it is most of what actually matters.

How should this be reported to a client?

By separating query types before showing the trend. A blended chart says traffic is down and invites the wrong conclusion. Split informational from transactional and the story becomes legible: one is being absorbed, the other is the business, and only one of them needs a response.

Search Console SEO AI 2026 Trends

Google Search Console: AI Overviews Data and What to Do About the Clicks You’re Losing

AI Overviews are eating informational-query clicks. Here’s the data and the playbook.

John Cravey with AIFounderApr 1, 20264 min readUpdated Jul 6, 2026

Google’s AI Overviews now appear on a meaningful share of informational queries. When AIO is present, the click-through rate to the underlying results drops significantly — the user gets the answer without clicking. GSC now exposes data about AIO impressions, and the data is sobering. Here’s what we’re seeing across the FH client book and the content adjustments we’ve made.

Free estimate · 2 minutes

Your Search Console data already shows the gap.

The estimate doubles as intake for an audit: we read your data, show where the clicks are being lost, and price the fix.

#Where AIO data appears in GSC

GSC’s Performance report shows ‘AI Overview’ as a SERP feature filter (alongside ‘Top stories,’ ‘Image pack,’ etc.). Filter by it to see queries where your URLs appeared inside an AIO panel — either as a source citation or as a related result.

Appearing in an AIO citation doesn’t mean the user clicked. CTR on AIO-cited results is dramatically lower than on traditional snippets at the same position. The data exists but the click-through reality is what matters.

#What we’re seeing across the FH client book

AIO appears on ~28% of informational queries our clients rank for.
AIO appears on <2% of transactional queries (‘[service] near me,’ ‘cost of [thing] in [city]’).
When AIO is present, average CTR drops 40-60% on the underlying organic results, even at position 1.
Blog content that targeted broad informational queries has seen the steepest click drops.

#Which queries get AIO

Mostly:

‘What is X’ and ‘how does X work’ explainer queries.
‘Best X for Y’ comparison queries where Google can synthesize from multiple sources.
Basic how-to queries with a definitive answer.
Definitional queries (‘meaning of X’).

Mostly NOT:

Local queries (‘plumber near me’).
Branded queries (someone searching your business by name).
Transactional queries with clear purchase intent.
Multi-step decision queries where users want to evaluate providers.
Highly specialized professional content where Google’s confidence is lower.

#The content strategy shift

If 28% of your traffic was coming from informational queries that now get AIO, you’re losing 12-17% of total clicks. The response is to shift content toward query types AIO doesn’t (or barely) cover.

Local content. Pages about specific cities, neighborhoods, regional concerns. AIO doesn’t synthesize local well.
Transactional content. Cost guides, pricing pages, comparison pages, ‘how to hire X’ guides where the user is making a decision.
Provider-specific content. Reviews, case studies, examples of your actual work. AIO can’t replace this.
Deep technical content where the user wants the expert, not a summary.
Process content. How you specifically do the work. AIO can summarize the topic but not your differentiation.

#Should you optimize content TO be cited in AIO?

Mixed. Being cited as a source in AIO gives you a link in the panel that may produce some clicks (typically 3-8% of AIO impressions click through to a source). But the user has already been answered, so the click is informational-only — they’re fact-checking. Brand credibility benefit exists; revenue benefit is marginal.

If you’re going to optimize for AIO citation, follow the same patterns Google rewards: clear, well-structured content; named author with expertise; recent updates; structured data (FAQ schema, HowTo schema where applicable); citations from other credible sources.

#What to STOP writing

Generic informational content that competes with AIO. Stop writing ‘what is SEO,’ ‘how does Google work,’ ‘what is a sitemap.’ Those queries get AIO panels with synthesized answers; clicks are minimal even at #1. The opportunity cost of writing them is real — that’s content effort that could go toward queries that still click.

#Repositioning existing informational content

If you have a library of informational content that’s lost clicks to AIO, the right move is usually not to delete it. It’s to convert it: add provider perspective, add local specificity, add comparison frameworks, add cost data. The same general topic with a strong point-of-view often still wins clicks because users want the expert take, not the synthesized take.

The traffic drop diagnostic post has more on triaging which content to update vs. delete.

#AIO impact on transactional content

Mostly minimal. Transactional queries (‘plumber Plano,’ ‘kitchen remodel cost Dallas,’ ‘best CRM for accountants’) still drive clicks at near-historical CTRs. AIO rarely fires on transactional intent. Local services SEO has been almost completely unaffected. Local SEO for construction firms is in a great spot — AIO doesn’t touch the queries that produce real leads.

#Measuring the AIO-attributed loss

Pull last 12 months of GSC clicks for your top informational queries.
Identify which queries now show AIO. Use GSC’s SERP feature filter or a SERP-checking tool.
Calculate the click delta on those queries from pre-AIO to post-AIO.
Compare to total click trend. The diff is your AIO-attributed loss.

#What good content for the post-AIO era looks like

Specific. Real numbers, real examples, real local detail. AIO doesn’t synthesize specifics well.
Opinionated. Take a position. AIO summarizes the consensus; differentiated content has the opinion.
Provider-perspective. Show how you specifically do the work. AIO can’t cite an experience.
Updated. Recently updated content gets more weight from Google and is more likely to be a cited source.
Structured. Use the FAQ, HowTo, Article schemas. Helps both AIO citation chance and traditional snippet quality.

#How this lands across FH client work

We’ve shifted every FH client content plan in the last 12 months. Less generic informational; more transactional, local, and provider-specific. The result on BHR Construction: organic clicks still up YoY despite AIO eating their informational tail. The transactional content has more than absorbed the loss. If your content plan still leans heavily on informational, book a consultation — the realignment is a one-quarter project that holds up.

Answers

Frequently asked questions

Where does AI Overview data appear in Search Console?: It is folded into your ordinary Performance data rather than broken out as its own report, with search-appearance filtering giving partial visibility. That means the impression you gained and the click you did not are in the same numbers as everything else, which is why the loss is easy to miss.
Which queries tend to get an AI answer?: Informational ones: definitions, how-to questions, comparisons, and anything with a short factual answer. Transactional and local queries are much less affected, because the searcher wants a provider rather than an explanation, and there is nothing for an answer to replace.
What does the pattern look like in the data?: Impressions holding or rising while clicks fall on informational queries, with position roughly unchanged. That combination is the signature. If clicks fell and impressions fell with them, something else happened and the answer feature is not your explanation.
Should I try to be cited in the AI answer?: Yes, and the work is the same work: direct answers near the top of the page, specific and sourced, on pages that are crawlable and fast. There is no separate optimization; being quotable is what earns the citation, and it also improves the page for humans.
What content should I stop writing?: Generic informational pieces whose whole value is explaining something an answer engine now explains in three sentences. That content had a traffic strategy behind it and the strategy has expired. The hours are better spent on pages that answer buying questions.
What should I do with existing informational content?: Reposition rather than delete. Add the specifics, the experience, and the position that a generated answer cannot supply, and connect it to the pages that convert. A generic explainer becomes useful again when it says something only your business can say.
Does this affect transactional content?: Much less. Queries where someone is choosing who to hire still send clicks, because an answer cannot make the decision for them. That asymmetry is the whole strategy: shift weight from explaining toward helping someone choose.
Is losing informational clicks actually bad for the business?: Usually less than the chart suggests. Traffic that arrived to read a definition rarely converted. The number to check is enquiries, not sessions, and a drop in one with the other flat is a reporting problem rather than a revenue one.
How do I separate this from a ranking drop?: Check whether position moved. An answer feature absorbing clicks leaves your position roughly where it was; a ranking drop moves it. Confusing the two sends teams rewriting pages that are still ranking exactly as well as they were.
Should I block AI features to protect my traffic?: You cannot block the answer feature without removing yourself from the search results it draws on, which trades a partial loss for a total one. For a business that wants customers, that is not a trade, and the option mainly exists for publishers with a different business model.
What is the practical response for a small business?: Concentrate on local and transactional queries, make every money page answer its question directly, and stop measuring success by informational sessions. That is a smaller change than the industry commentary suggests and it is most of what actually matters.
How should this be reported to a client?: By separating query types before showing the trend. A blended chart says traffic is down and invites the wrong conclusion. Split informational from transactional and the story becomes legible: one is being absorbed, the other is the business, and only one of them needs a response.

Question we did not answer? Ask us directly and we will answer it here.

Tagged#Search Console #SEO #AI #2026 Trends

Written by

John Cravey

Founder

Founder of Frontend Horizon. Writes most of the long-form work on the FH blog.