AI Overviews and SEO: The Data on Click Behaviour

Qualified traffic vanishes whenever Google answers a query and the user never clicks.

This playbook covers how to regain visibility by earning citations in AI Overviews, with a method built on data, fixes and testing. Within a day you can audit Search Console for the affected queries, confirm your pages are crawlable, and check GA4 for what happens after the visit.

At a Glance

What AI Overviews SEO is in 2026

Succeeding here means formatting content so Google treats it as a primary source, which protects organic CTR as the results page changes shape.

Google assembles answers from multiple domains at the top of results, often bypassing the traditional list. Some searchers get what they need without clicking, but the source cards remain a door to your site for anyone who wants depth.

The scale of the shift is worth being accurate about. Pew's analysis of 68,879 searches found 18% produced an AI summary — significant, and not a replacement for classic search. Sessions also ended more often on visits with a summary, 26% against 16%.

You'll also see Gartner's February 2024 forecast that search volume would fall 25% by 2026 quoted everywhere as settled fact. It was a forecast, we're now in the year it referred to, and Google adapted by building AI Overviews into Search while retaining its market position. Treat it as directional.

Why AI Overviews are changing organic CTR

AI Overviews create a zero-click dynamic. Synthesised answers with source cards mean fewer people click through.

The pattern is uneven, and that's useful. Pages serving multi-step informational queries see the largest CTR drops. Transactional and brand-specific searches hold up better, because those users intend to evaluate or buy rather than learn. Detailed how-to guides and definitions lose the most.

Track clicks and on-page engagement together, so you can tell whether a citation is building brand awareness or simply absorbing your value.

• Check Search Console for long-tail queries with sudden CTR declines
• Build a landing-page cohort in GA4 for those pages, tracking engagement and conversions after the shift
• Prioritise pages where high-intent actions started with organic visits, not low-value listicles

How Google decides what goes into AI Overviews

Google weighs query complexity, index status, content clarity and site authority.

Complex questions trigger a fan-out process: the query decomposes into sub-questions, multiple pages are retrieved for each, and the best parts are synthesised with source cards attached.

Technical eligibility comes first. Pages must be indexed and not blocked by robots.txt or noindex. Then clarity — one topic per page, facts stated plainly. Authority is the deciding factor between comparable pages, which is where backlinks and brand mentions do their work.

One thing worth stating because it's commonly confused: Google-Extended controls whether your content is used to improve Gemini. It does not control AI Overviews eligibility, which draws on the standard Search index. Blocking Googlebot removes you from Search entirely.

How to optimise for AI Overview citations

Three areas: technical stability, matching user intent, and external proof.

There's research behind the content side. The 2023 paper that coined generative engine optimization tested strategies across 10,000 queries and found adding statistics improved visibility by around 41%, and adding quotations by around 28%. The largest gains went to pages ranking around position five, where citing external sources improved visibility by up to 115% — while pages already at position one saw little change.

That's a prioritisation answer. Mid-ranking pages have the most to gain.

Start with a full audit and prioritise pages that generate leads or demos. Scattering effort across everything makes the impact impossible to read.

Getting crawled and indexed

• Check robots.txt, noindex tags, sitemaps and canonicals
• Run a site audit for 4XX and 5XX errors
• Confirm index status in Search Console

Target long-tail queries

• Mine Search Console for the specific questions and phrases resembling People Also Ask results
• Filter for long-tail phrases showing buying intent
• Check GA4 for which long-tail searches historically sent traffic

Structured data and citations

• Use Article, FAQPage or Person schema to define authorship and page type
• Include citations and direct links to sources in your text
• Validate with the Rich Results Test and watch Search Console for markup errors

Be precise about what schema does. Google's structured data documentation specifies no AI-specific markup, and structured data is not a requirement for appearing in AI Overviews. It clarifies your facts for machines rather than qualifying you for the feature.

Content quality and E-E-A-T

E-E-A-T covers Experience, Expertise, Authoritativeness and Trustworthiness. Worth knowing what it actually is: Google's guidance on helpful content describes it as a framework their quality raters use to evaluate whether ranking systems are working. It isn't a score you optimise directly.

Demonstrate experience with real examples, your own data, screenshots of work. Establish expertise through detailed author bios and links to original research. Build trust with accurate sourcing, visible update dates and transparent corrections.

• A case study with specific numbers, dates and the tools actually used
• An author bio with a professional title, links to original work, and third-party mentions

Backlinks and brand mentions

External signals remain a major factor. Links and mentions tell search systems your content is a source worth trusting.

• Fix broken links and pitch for inclusion in niche resource lists
• Use PR to earn mentions in publications your buyers read
• Track progress with backlink tools and brand alerts on G2 and LinkedIn

How to measure the impact

Combining Search Console and GA4 shows whether a summary helps or siphons traffic. Impressions alone mislead.

An important limitation to plan around. Search Console does not report AI Overview appearances separately. Google includes clicks and impressions from AI Overviews within overall Search performance data, with no filter isolating them. Anyone telling you to filter for generative features is describing something that doesn't exist.

What you can do is infer. Identify queries where impressions hold steady while CTR falls sharply — that pattern is consistent with an overview appearing above your result.

Measurement checklist

• Export Search Console query data and look for impressions-flat, CTR-falling patterns
• Build GA4 landing-page segments to compare engagement and conversions before and after the shift
• Tie content refreshes to dates so you can isolate why performance moved

GA4 behaviour metrics

• Sessions and engagement time on pages that appear in overviews
• Landing-page segments comparing behaviour across exposure windows
• Conversions linked to those segments, to see whether remaining traffic still converts

Building a human-in-the-loop workflow

The agent proposes, you decide. Automation handles the scale of finding problems; you own accuracy and brand voice.

1. A signal flags a CTR drop and lists the affected pages
2. Your content marketer reviews the proposed changes in HubSpot
3. They approve, amend or reject
4. Approved updates ship with a date and a change record
5. You track recovery in Search Console and GA4

When CTR drops, it becomes a task. The marketer checks the page, confirms facts, and verifies the author bio holds up. Your governance plan needs set review times and a rollback path if conversions fall. More on human-in-the-loop here.

Where Strivelabs fits

Strivelabs pulls GA4, Search Console, Google Ads and HubSpot together and flags the pattern this article is about — a page holding impressions while losing clicks, particularly one you're also paying to send traffic to.

It gathers the affected queries, proposes an update, and drafts a task in HubSpot with a named owner. A marketer reviews before anything ships, and approval is enforced by permission rather than convention. Afterwards it watches Search Console and GA4 to see whether the page recovered.

The honest limitation: it won't write the update or decide your positioning. Those stay judgement calls.

If CTR drops are going unnoticed until someone happens to look, book a demo.

Conclusion

AI Overviews SEO is now a core part of B2B strategy, and the plan works best kept simple. Target technical health, answer specific long-tail questions properly, add structured data and real author credentials, and track results in Search Console and GA4.

Start with your top pages. Where CTR keeps falling on specific queries, treat it as a task rather than a trend. Update the page and the schema, then watch citations and CTR for a fortnight.

Frequently Asked Questions