Introduction
For about eighteen months, the conversation about Google AI Overviews has been almost uniformly grim. Pages dropped clicks, panels got truncated, and a generation of small-business owners reasonably wondered whether organic search was still a place worth investing. The new data tells a more complicated story — one that should change what you do next quarter, but not in the direction most agencies are about to recommend.
A study published this week by Seer Interactive, covered by Search Engine Land on April 24, looked at 53 brands across 5.47 million queries and 2.43 billion impressions from January 2025 through February 2026. Two findings stand out. First, click-through rates on pages where AI Overviews appear have begun to recover from their late-2025 lows. Second, the recovery is uneven across query types in ways that are useful — and slightly counterintuitive — for small businesses to plan around.
This post unpacks the numbers, explains the mechanisms behind the recovery, and walks through a concrete 30-minute Search Console workflow you can run tomorrow to measure whether your own site is participating. We will also be honest about what the study does not tell us, where the panic responses of 2025 were over-corrections, and where they were not.
Key Takeaways
- Seer Interactive's April 2026 study of 5.47 million queries found AI Overviews CTR rose from 1.3% in December 2025 to 2.4% in February 2026 — meaningful recovery, but still well below the 3.8% CTR on results without an AI Overview present
- Pages cited inside an AI Overview earn roughly 2.1% CTR, while non-cited pages on the same SERP earn about 0.9% — citation, not just ranking, is the variable that matters
- Recovery is fastest on comparison and question-format queries (where AI Overviews appear ~95% and ~86% of the time) and least relevant on transactional queries (~5%)
- The 2025 wave of “rewrite everything as FAQs” was an over-correction; the 2026 response looks different — long-form for commercial intent, citation-friendly structure, and proprietary data
- A 30-minute Search Console diagnostic compares your CTR delta to the study averages and tells you whether you are participating in the recovery
- AI Mode and continued query-shape changes mean the recovery may not be the final equilibrium — plan for another reset within 12–18 months
What Did the New Study Actually Find?
The headline number is a recovery, but the recovery is bounded and the bounds matter.
According to the Seer Interactive analysis covered by Search Engine Land, aggregate organic click-through rate on results pages where AI Overviews appeared rose from 1.3% in December 2025 to 2.4% in February 2026 — described in the article as an 85% jump. Over the same window, CTR on results without an AI Overview present rose from roughly 2.8% to 3.8%, about a 35% increase. The gap between the two states narrowed but did not close. SERPs with AI Overviews still produce meaningfully fewer clicks than SERPs without them.
The more useful number is buried below the headline. Pages that are cited inside the AI Overview earn an average CTR of roughly 2.1%, while pages on the same results page that are not cited earn around 0.9%. As the article puts it, “Pages cited in [AI Overviews] get more clicks than pages on that same results page that aren't cited” — but those cited pages still receive fewer clicks than pages on AIO-free SERPs. In plain English: being cited is a meaningful improvement over being uncited on an AIO page, but the safest place to be is still a query that does not trigger an AI Overview at all.
That last sentence is where small-business strategy gets interesting, because query type controls AIO appearance rates. Per the same Seer dataset, comparison queries trigger AI Overviews about 95% of the time, question-format queries about 86% of the time, informational queries about 36%, and transactional queries only about 5%. Paid search numbers move in a related but distinct pattern — paid CTR on AIO results rose from 14.6% to 16.2%, while paid CTR on non-AIO results actually fell from 26% to 21.8%. We unpacked the paid-side mechanics in how AI Overviews are reshaping paid search; this study refines that earlier picture rather than replacing it.
For context on how large the original drop was, an earlier Search Engine Land piece from April 2 on why content doesn't appear in AI Overviews cited Seer Interactive data showing informational-query CTR fell 61%, from 1.76% to 0.61%, when AI Overviews were present. We covered the inclusion-criteria angle for small businesses in our companion post on why content doesn't appear in AI Overviews even when it ranks. The recovery to 2.4% is real, but it is a recovery to a new equilibrium, not to pre-AIO baselines.
Why Is CTR Recovering at All?
Three mechanisms are doing most of the work, and only one of them is something Google chose.
The first is user adaptation. As AI Overviews became universal across query types, users learned that the AI summary at the top of the page is generated, sometimes wrong, and rarely sufficient for a real decision. The behavior we are seeing — and the BrightEdge research summarized by Search Engine Land supports this — is a partial return to verification clicking. Users skim the AIO, then click through to a cited source to confirm. That is exactly what cited-page CTR of 2.1% is measuring.
The second is the slow shift in citation density. Google's AIO panels in early 2025 often summarized without citing visibly, or cited sources in ways users did not register. The current panels make citations more legible, with discrete source cards that read as click invitations rather than footnotes. This is a Google-side product change that quietly favors recovery without Google ever announcing it.
The third is query shape, and this is where Google's Liz Reid is most useful. In comments published April 23 in Search Engine Land, Reid noted that AI Overviews have driven “meaningfully longer queries” and “more natural language queries.” She added that “people stop talking just in keywordese as much, and they start expressing more of what they want.” Longer, more natural-language queries skew toward exactly the comparison and question intents where AI Overviews now appear most often — which is also where citation-pattern recovery is strongest. Reid's observation that “before AI slop, there was slop” is also worth keeping; the underlying quality contest has not changed, just the surface.
According to the same April 2 Search Engine Land coverage of Pew Research Center data, only about 8% of users who encountered an AI Overview clicked through to a traditional search result, compared to 15% on AIO-free SERPs. The Pew sample was earlier in the cycle, so the gap may be narrowing today, but the structural point holds: AI Overviews compress the click market, and recovery is happening inside that compressed market, not by reversing the compression.
Where the 2025 Panic Responses Were Over-Corrections
This is where we have to be uncomfortable, because some of the standard 2025 advice — including advice we and others gave — looks like an over-correction in light of the new data.
The biggest one is the “rewrite every page as a FAQ” move. The intent was sound: make content easier for AI Overviews to extract. But the recovery data suggests that question-format queries are exactly where AIO appearance is highest (around 86%) and where CTR compression is most severe. Aggressively reformatting commercial-intent pages into FAQ structures pushed those pages into a higher-AIO bucket without meaningfully increasing citation rate. Some of those rewrites were a tax, not an upgrade.
A related over-correction was snippet-shortening. The 2025 advice was to compress every page's primary answer into the first 60 words to maximize extractability. That worked for pure information queries, but for products, services, and comparisons — the queries that actually drive small-business revenue — it stripped out the depth that made the page worth visiting in the first place. The recovery data shows that cited pages still earn 2.1% CTR; uncited pages earn 0.9%. Being short and uncited is worse than being long and cited.
The third over-correction was abandoning long-form. We have seen Fort Wayne and Northeast Indiana clients quietly stop publishing 2,500–3,000-word pieces last year because “AIO eats the traffic.” The new data does not support that retreat for commercial-intent and bottom-funnel content, where AI Overviews appear about 5% of the time on transactional queries. This is also why our own bottom-funnel content for AI search playbook has held up better than mid-funnel content over the same window — it sits in the part of the SERP where AIO compression is least active.
To be fair to the 2025 advice: AIO impact was still being measured and the safe move was defensive. But it is now visible that some of the defensive moves were costlier than the original problem.
What the 2026 Response Actually Looks Like
Five concrete adjustments fit the recovery-era data. None of them are dramatic; the dramatic-sounding moves are mostly what we are now backing away from.
Keep long-form for commercial intent and product comparisons. The Seer data shows transactional queries trigger AIO only about 5% of the time. For bottom-of-funnel content, you are competing in mostly-classic SERP territory, where depth, internal linking, and conversion design still drive performance. The pillar-page playbook is not dead; it is just not the right shape for everything.
Pair short answers with detailed methodology. Where AI Overviews appear, the structure that gets cited is a clear, self-contained answer near the top of the page paired with a transparent, sourced methodology section below. The 60-word answer at the top earns the citation; the methodology earns the click-through. That is the structure our Answer Engine Optimization guide walks through in detail and the structure the citation data continues to reward.
Prioritize first-party data and proprietary numbers. AI Overviews synthesize across sources well; they cannot synthesize numbers that exist on only one site. A small business that publishes its own service-area cost benchmarks, customer-survey results, or operational data has content that the AI cannot reproduce and therefore must cite. The ChatGPT citations favor ranking and precision analysis we ran earlier this year backs this up — proprietary data shows up disproportionately in citation slots.
Lean into commercial-intent and bottom-funnel content. The transactional-query AIO appearance rate of about 5% is where you want to live if AIO compression is the threat you are most worried about. Service pages, location pages, comparison pages between specific products, and pricing pages all sit in low-AIO territory. We have written about this at length in our bottom-funnel content for AI search playbook.
Build click-worthy SERP entries, not answer-aggregator entries. When your page is competing on a question-format query that does trigger an AIO, the goal shifts. You are no longer competing for “did the user find the answer.” You are competing for “did the user, having seen the AIO answer, decide your page is worth opening for verification or depth.” That is a different copy task — closer to display-ad copy than classic SEO meta-description writing — and it is one we are still learning to do well.
A 30-Minute Search Console Recovery Diagnostic
Here is the workflow you can run tonight to measure whether your own site is participating in the recovery the Seer study describes. It uses only Google Search Console and assumes you have at least 90 days of impression history.
Step 1 — Pull the comparison window. In Search Console, go to Performance → Search results. Set the date range to “Compare,” with one period being the last 30 days and the comparison period being December 2025 (Dec 1–31). Export the page-level data with both impression and CTR columns.
Step 2 — Segment by query type. This is the step most owners skip. Use the Queries tab and filter for three groups. For comparison queries, filter “Query contains” with words like vs, versus, or, compare, better than, cheapest. For question queries, filter “Query contains” what, how, why, when, where, should, can, do. For transactional queries, filter “Query contains” near me, price, cost, buy, quote, hire, book. Each filter group will export a different CTR delta.
Step 3 — Calculate per-segment CTR change. For each of the three filter groups, compute the CTR delta from December 2025 to last 30 days. Compare to the Seer benchmark of about an 85% rise on AIO-present queries and 35% rise on non-AIO queries. Your numbers will probably not match — they are individual-site numbers, not aggregate market numbers — but the direction is what matters.
Step 4 — Flag pages where impressions are up and CTR is down. This is the “you are showing up in AIO but not getting cited” pattern. Sort the page-level export by impression growth, then look at CTR change for the top 20 growth pages. Pages where impressions grew but CTR fell are the priority candidates for the citation-friendly restructuring above.
Step 5 — Tag pages where CTR is recovering. These are the pages where the structure already works. Reverse-engineer what they have in common: structure of the first 100 words, presence of a sourced methodology block, internal-link density, presence of original data, schema markup. Whatever they share is the template you replicate to the pages that are not recovering.
The full pull, segmentation, and analysis takes 30–45 minutes if you have done a Search Console export before. We covered the deeper version of this approach in our intent gap analysis in Search Console playbook — the recovery diagnostic is a faster, narrower version of the same exercise.
Honest Caveats
A few things the Seer study does not tell us, and a few it does that should be qualified.
Single-study limitation. This is one analysis of 53 brands. The brands appear to be Seer Interactive clients or partners, which biases toward larger, better-instrumented sites. A small Northeast Indiana service business operating on a few thousand monthly impressions will not see the same shape of recovery; the variance at low impression counts is too high.
Geographic and language scope. The study is implicitly U.S. English. We have no recent data on whether AIO recovery looks similar in Spanish, French, or German markets, or in U.S. local-pack queries where the AI integration is different.
Recovery may level off. The 1.3% to 2.4% jump happened in two months. There is no guarantee the line keeps climbing. It may be approaching a new equilibrium below the 3.8% non-AIO baseline. Plan for the new equilibrium, not for a return to 2024 numbers.
AI Mode resets the clock. Google's Liz Reid commentary in Search Engine Land flagged that query shapes continue to evolve and that AI Mode in particular is a different surface again. The recovery measured in the Seer study is for AIO panels inside classic SERPs. AI Mode behaves differently and may reset the click compression once it reaches the same scale.
Causation is uncertain. The study shows correlation — CTR rose. The three mechanisms we cited above (user adaptation, citation density, query shape) are plausible but not proven. Other explanations exist, including selection effects in the brand sample.
This is why we tell clients to run their own Search Console diagnostic rather than rely on the aggregate number. The aggregate number is directionally useful; your number is what you act on.
What to Monitor Over the Next 90 Days
If you are going to track one thing, track per-segment CTR delta — the same three segments from the diagnostic above. Beyond that, four metrics will tell you whether your site is participating in the recovery or being left behind.
First, citation appearance rate. Run 20–30 of your highest-priority queries through Google AI Mode and ChatGPT once a week and record whether your domain appears as a cited source. The rate of change matters more than the absolute number.
Second, impressions on AIO-triggering queries. If impressions on comparison and question queries are flat or rising, AIO is showing your page; the question becomes whether it is citing it.
Third, click-through-rate on cited pages versus non-cited pages on your own site. The Seer aggregate is 2.1% versus 0.9%. If your gap is wider than that, your cited pages are working harder than the market average, which is the cleanest signal that the citation-friendly restructuring is paying.
Fourth, conversion rate on AIO-driven sessions. Recovery in clicks is good; recovery in clicks that do not convert is a different problem. Tag AIO-cited landing pages in your analytics and watch for a conversion-rate gap. We see one regularly, and it usually traces to the page being optimized for citation extraction at the cost of in-page conversion design.
Where This Fits in Our Work
For Northeast Indiana small businesses, the practical question is whether to invest in a structural rewrite of your top-30 pages this quarter or wait for more data. Our default answer for clients is: do the diagnostic, then do the highest-leverage 5–10 pages first, then watch the per-segment CTR for 60 days before scaling. That is conservative on purpose. The recovery is real but bounded, and the worst version of any AIO response is one that turns a content portfolio sideways inside a 30-day sprint based on a single study.
If you would like to walk through your own diagnostic with us, our Answer Engine Optimization service and SEO service both include a quarterly AIO citation review as standing work. If you ran the panic responses last year, the more useful conversation right now is which of them to undo — that is often a faster lift than starting fresh. For broader local-pack context, see also where the traffic actually went for Northeast Indiana businesses through 2025.
Run the AIO Recovery Diagnostic on Your Site
Button Block runs the 30-minute Search Console diagnostic, the per-segment CTR review, and the citation-friendly restructuring as standing work for Northeast Indiana clients. We can walk through your data with you and show which 5–10 pages are the highest-leverage starting point.
Frequently Asked Questions
- Is the AI Overviews CTR recovery real or just one study?
- The recovery is currently documented in one study — Seer Interactive's April 2026 analysis of 53 brands, reported by Search Engine Land. A single study is suggestive, not definitive. The numbers (CTR rising from 1.3% to 2.4% on AIO-present SERPs from December 2025 to February 2026) are large enough and the sample is broad enough to take seriously, but you should run your own Search Console diagnostic before assuming your site participates in the recovery at the same rate.
- Should I undo the FAQ rewrites I did in 2025?
- Not all of them, but probably some. FAQ structure is genuinely useful for question-format queries that trigger AI Overviews about 86% of the time. It becomes a tax when applied to commercial-intent or comparison pages where the user needs depth, not extractability. Pull a Search Console export, identify any commercial-intent pages where you converted to FAQ structure last year, and check whether their CTR has recovered. If not, restoring depth and adding a sourced methodology block is usually the right move.
- What is the difference between being cited in AI Overviews and being ranked?
- Citation is which sources the AI Overview links to inside the panel. Ranking is your position in the classic blue-link results below the panel. The Seer data shows pages cited inside AIO earn about 2.1% CTR, while non-cited pages on the same SERP earn about 0.9% — being cited matters more than being well-ranked when an AIO is present.
- Which queries are still safe from AI Overview compression?
- Per the Seer dataset, transactional queries trigger AI Overviews only about 5% of the time. Pricing pages, "near me" searches, and direct service queries ("hvac repair in [city]") sit largely outside the AIO compression zone. Comparison queries (95% AIO appearance) and question queries (86%) are the most exposed.
- Will AI Mode reset the recovery?
- It might. Google's Liz Reid has signaled that AI Mode behaves as a different surface from AIO panels inside classic SERPs. If AI Mode reaches similar query share, the click compression we saw in 2024–2025 could repeat there, with another recovery cycle to follow. Plan for the next reset within 12–18 months rather than treating the current recovery as a permanent state.
- How do I know if my own site is participating in the recovery?
- Run the Search Console diagnostic in this post: pull a 30-day-vs-December comparison window, segment by query type using filters for comparison, question, and transactional intent, and calculate the CTR delta per segment. If your AIO-present queries show CTR rising at all, you are participating; if they are flat while non-AIO queries are rising, you are still being compressed and the citation-friendly restructuring above is the next move.
- Does the recovery apply to local search and the local pack?
- We do not yet know. The Seer dataset is general U.S. organic results. Local-pack queries and Google Maps queries have a different AI integration, and the traffic-loss patterns we documented earlier for local Northeast Indiana businesses operate through different mechanics than the AIO panels above the organic blue links. Treat the Seer numbers as a national organic-search benchmark, not a local-search forecast.
Sources & Further Reading
- Search Engine Land: searchengineland.com/google-ai-overviews-ctr-recovery-study-475566 — AI Overviews and click-through rate: New Seer data shows recovery (April 24, 2026)
- Search Engine Land: searchengineland.com/why-content-doesnt-appear-in-ai-overviews-473325 — Why your content doesn't appear in AI Overviews (April 2, 2026)
- Search Engine Land: searchengineland.com/google-liz-reid-ai-search-query-shifts-ai-slop-475324 — Google's Liz Reid on AI search query shifts and AI slop (April 23, 2026)
- Neil Patel: neilpatel.com/blog/ai-paid-search-adapt-to-ai-overviews-stealing-clicks — How paid search teams should adapt to AI Overviews (April 9, 2026)
- Seer Interactive: seerinteractive.com/insights — Seer Interactive research insights on AI search and CTR
- BrightEdge: brightedge.com/news/press-releases — BrightEdge AI search research and press releases
- Pew Research Center: pewresearch.org/short-reads/2025/03 — Americans and AI in search behavior data
- Google: support.google.com/webmasters — Google Search Console help documentation