Your Search Console impressions chart is climbing. Your clicks chart is flat, or worse, sliding. Average position hasn’t moved. If you lead marketing or growth at a SaaS company and you’ve stared at this exact combination of lines wondering whether your rankings are actually fine or quietly dying, you are looking at one of the most common and most misdiagnosed patterns in 2026 organic search: impressions inflate while clicks stagnate because a growing share of your visibility is happening inside an AI Overview, not on your listing.
This is not a ranking problem. It is a visibility-versus-capture problem, and the two require completely different fixes. Treating it as a ranking problem, by publishing more content or chasing new keywords, usually makes the pattern worse, because more indexed pages means more surface area for AI Overviews to summarize without sending a click.
Below is a five-step diagnostic framework for telling AI Overview cannibalization apart from the other three things that produce the same chart shape, plus how to combine Google Search Console with GA4 to confirm what’s actually happening before you touch a single page. If you need a refresher on what impressions actually measure in this context, our glossary covers the definition in more depth.
What the data actually shows about AI Overviews and clicks
Seer Interactive’s ongoing tracking study, which follows 5.47 million distinct queries and 2.43 billion organic impressions across 53 brands from January 2025 through February 2026, found that organic CTR on AI-Overview-present queries fell from 3.19% in January 2025 to 1.31% by December 2025, then partially recovered to 2.36% by February 2026 as searchers adapted to the new SERP layout. Their per-million-impression breakdown is the more useful number for diagnosis: queries with no AI Overview generate roughly 33,500 organic clicks per million impressions, queries with an AI Overview that cites your domain generate about 20,743, and queries with an AI Overview that does not cite you generate about 9,445. Being cited inside the Overview roughly doubles your click yield compared to not being cited, but both are well below what the same query produced before Overviews existed on it. Source: Seer Interactive, AIO Impact on Google CTR: 2026 Update.
Google itself effectively confirmed how much this pattern was confusing site owners when it rolled out a branded-versus-non-branded query filter directly inside Search Console, announced in November 2025 specifically to help teams separate demand-side changes (fewer people searching your brand) from capture-side changes (the same people searching, but Google answering them without a click). Source: Google Search Central, Introducing the Branded Queries Filter in Search Console, also covered independently by Search Engine Land.
That filter is a useful ingredient, but on its own it does not tell you why non-branded clicks fell. It only tells you whether the drop is on the brand or non-brand side. You still need a process to get from “non-branded clicks are down” to “here is specifically why, and here is what to do about it.”
The Capture Gap Framework: five steps to diagnose an impressions-up, clicks-flat pattern
This is an original synthesis built for SaaS marketing teams, not a restatement of any single source above. It combines the branded query filter, page- and query-level Search Console data, and GA4 landing page data into one sequence, because none of those three data sources alone is conclusive.
Isolate the pattern with the branded query filter
In Search Console’s Performance report, apply the branded queries filter and compare branded versus non-branded trends over the same 16-week window. If branded clicks are stable and non-branded clicks are flat or falling while non-branded impressions climb, you have ruled out a brand-demand problem and a broad algorithmic demotion. That combination is the specific signature of AI Overview capture, not a ranking loss.
Pull the query list and check average position against CTR, not clicks
Export non-branded queries and sort by impression growth. For each, look at average position and CTR side by side. Queries holding position 3 to 6 with CTR under 1 percent are the tell. If position held steady while CTR compressed, ranking is not the issue. If position also dropped, you have a genuine ranking problem layered on top, and you should fix that first.
Spot-check the SERP for an AI Overview on your flagged queries
Manually search the top 15 to 20 flagged queries in an incognito browser. Note whether an AI Overview appears, and whether your domain is cited inside it. This step cannot be fully automated from the API today, but 15 to 20 manual checks is enough to confirm or reject the hypothesis for a query set.
Cross-reference against GA4 landing page sessions for the same URLs
Pull Organic Search sessions by landing page in GA4 for the same date range and match them to the flagged Search Console URLs. Remember these two tools do not measure identically: GSC counts a click any time your listing is clicked from Google Search or Google Discover, while GA4’s Organic Search channel groups sessions across all recognized search engines and can undercount when consent mode limits tracking. Expect the two to be directionally close, not to reconcile exactly. If GA4 sessions for a URL are falling faster than GSC clicks, check for a tracking or consent issue before blaming search.
Split the fix by whether you’re cited in the Overview
If you are already cited, the play is defensive: keep the page eligible by maintaining clear, extractable answer text near the top and current FAQPage schema, since Seer’s data shows cited pages retain roughly double the click yield of uncited ones. If you are not cited, the play is offensive: tighten the direct-answer paragraph to the exact question pattern, add or refresh schema, and check whether a competitor is winning the citation with a more extractable structure on an equivalent page.
Steps 1 and 2 alone will resolve most false alarms. Teams often skip straight to “rewrite the content” when the real finding at step 2 is that position dropped two spots during a core update, which is a ranking problem the Capture Gap Framework correctly routes away from an AI Overview fix.
How the four causes compare
Before assuming AI Overview cannibalization, rule out the other three explanations that produce a similar-looking chart. This site’s own breakdown of what average position in GSC actually measures is a useful companion read here, since average position is the metric most likely to mislead you at step 2.
| Cause | Impressions | Avg. position | Branded vs. non-branded | Fix |
|---|---|---|---|---|
| AI Overview capture | Up or flat | Stable | Non-branded only | Win the citation or tighten extractable answers |
| Ranking loss | Flat or down | Worsening | Both, or non-branded | Standard SEO remediation of the page or site |
| Demand decline | Down | Stable | Branded and non-branded both fall | Demand generation, not on-page SEO |
| Tracking or consent gap | Stable in GSC | Stable | GA4 diverges from GSC sharply | Audit GA4 tagging and consent mode config |
What this looks like in a real reporting view
The mockup below illustrates the pattern this framework is built to catch. It uses invented, illustrative figures only, not real client or company data, to show what a Capture Gap review looks like once GSC and GA4 are laid side by side for a flagged query set.
Illustrative dashboard mockup built for this article. All figures are invented placeholder numbers, not real client, account, or saasseo.com data.
Where the missing clicks sometimes actually go
One more check worth running before closing out a Capture Gap review: look at your GA4 referral traffic, not just Organic Search, for growth from AI answer engines. A client we work with in the education vertical saw ChatGPT referral sessions in that same GA4 property grow 87% year over year as of May 2026, with Gemini referrals up 135% and Claude referrals up 577% over the same period. That is a different client’s traffic, not saasseo.com’s own, but it illustrates the broader pattern worth checking in your own property: some of what leaves the Organic Search channel is not lost, it shows up later as a small but growing direct or referral line from an AI product, once someone acts on an answer they read elsewhere.
This is also where thin, generic pages hurt you twice. If your glossary or resource pages are shallow, AI systems have less reason to cite you at all. Our own SaaS marketing glossary now runs 342 substantive term definitions specifically because thin glossary entries were an early symptom of this same capture problem, both in traditional search and in AI citation behavior. If you’re building out this kind of GSC and GA4 review as a recurring process, it also connects directly to the crawl and indexing hygiene we covered in our llms.txt decision framework, since both are ultimately about controlling what gets surfaced to an AI system and whether you get credited for it.
FAQ
Does an AI Overview appearing on a query always reduce my clicks?
Not always, but it usually reduces click yield relative to the same query without an Overview. Seer’s 2026 data shows even cited pages retain only about 62 percent of the click yield of a no-Overview result, and uncited pages retain closer to 28 percent.
Can I tell from the Search Console API alone whether a query triggered an AI Overview?
No. Search Console does not currently expose an AI Overview presence flag at the query or page level through the API or UI. You have to spot-check manually, which is why step 3 of the Capture Gap Framework is a manual SERP check rather than an automated pull.
Why don’t my GA4 organic sessions match my GSC clicks exactly?
They measure different things. GSC counts clicks from Google Search and Discover specifically. GA4’s Organic Search channel groups sessions from any recognized search engine and depends on client-side tagging and consent state. Expect them to move together directionally, not to reconcile to the exact number.
Should I stop publishing content if I see this pattern?
No. Stopping publication does not fix a capture problem, it just slows your growth in query coverage. The fix is tightening the pages you already have to be more extractable and more likely to be cited, alongside your existing content plan, not instead of it.
If your team is seeing this pattern and needs a second set of eyes on whether it’s an Overview capture issue or something else in the query data, our technical SEO team runs this exact review as part of ongoing SaaS SEO engagements. Book a strategy call and bring your Search Console export.
