AI Overviews CTR: First Recovery After 12 Months of Decline
Summarize this article with AI
The Seer Interactive Study: 2.43 Billion Impressions, a Documented Turning Point
A client calls me last Tuesday. Outdoor equipment e-commerce, 12,000 organic sessions per month. He tells me: « Stéphane, we’ve lost 30% of traffic on our guides since May 2025. AI Overviews are killing us. »
I ask if he’s cited in the AIOs. He doesn’t know.
End of conversation. Start of the audit.
Two days later, Search Engine Land publishes the Seer Interactive study. 53 brands tracked. 5.47 million queries. 2.43 billion impressions analyzed from January 2025 to February 2026. First public dataset documenting a reversal in the trend.
The number that matters: average organic CTR on searches displaying an AI Overview has risen from 1.3% in December 2025 to 2.4% in February 2026. That’s +85% in two months.
Not an opinion. A measurement.
It doesn’t mean AI Overviews are no longer cannibalizing traffic. It means cannibalisation is no longer systematic. It’s become selective. It sorts. It punishes those not cited. It rewards those who are.
The question is no longer « are AIOs taking my traffic? ». The question is: « am I cited in the AIO or not? »
Because according to the same study, pages cited in an AI Overview capture 2.1% average CTR. Pages visible on the same SERP but not cited in the AIO drop to 0.9%. That’s 57% fewer clicks.
Conversely, searches that don’t display an AI Overview climb to 3.8% CTR in February 2026, versus 2.8% in January 2025. +36%.
Google isn’t diluting everything. It’s concentrating. It’s directing. It favors transactional and complex queries, it answers simple questions directly.
Where Do AI Overviews Actually Appear?
Not all query types are equal when facing AIOs.
The Seer Interactive study measures the frequency of AI Overviews appearing by search intent:
- Informational queries: 36% AIO appearance
- Transactional queries: 5%
- Comparison queries: 95%
- Question format queries: 86%
An e-commerce site selling trail shoes will almost never see an AIO on « buy women’s trail shoes ». Google shows ads and product cards.
But on « difference between trail and road shoes », the AIO appears 95 times out of 100.
On « how to choose trail shoes for beginners », there’s an 86% probability.
What does that mean for content architecture?
It means your transactional pages are protected. Your product cards, your catégories, your PPC landing pages aren’t threatened by AIOs. They continue capturing clicks.
What’s threatened is non-differentiating informational content. Generic guides. FAQs answering simple questions. Articles like « what is X » with no expertise or proprietary data.
Google no longer needs you to explain « a trail shoe has deeper lugs than a road shoe ». He’s absorbed that answer 10,000 times. He restates it in the AIO. Without citing you.
But if your guide contains a comparison table of 15 models, with field tests, measured durability, foam density graphs — Google cites you. Because it can’t create that data. It can only reference it.
I see this every day with my clients. The pages appearing in AIOs are the ones bringing primary data. The others sink.
Cited in the AIO: +133% CTR vs. Non-Cited Pages
Seer Interactive’s numbers are clear.
On a SERP with AI Overview:
- Page cited in the AIO: 2.1% CTR
- Page not cited but present in organic results: 0.9% CTR
That’s a +133% gap in favor of cited pages.
On a SERP without AI Overview:
- Average CTR: 3.8% (February 2026)
What does that mean concretely?
It means if you’re not cited in the AIO, you lose 57% of your potential CTR compared to a cited page. And you lose 76% compared to a SERP without an AIO.
Real client example. Industrial furniture brand. 847 content pages. Catalog structured in semantic clusters since November 2023. We track 1,200 queries.
In May 2025, Google starts massively rolling out AIOs on their comparison queries (« difference high table low table industrial », « what height for bar stool 90 cm », etc.).
Between May and December 2025: -22% organic traffic on the guides. Not on the product catalog. Just the guides.
We audit. We look at the AIOs. 68% of guides are never cited in the AI Overviews appearing on their target queries.
We restructure. We add dimension tables, annotated schemas, chiffered comparisons (weight, load capacity, height standard vs custom). We republish 40 guides between January and March 2026.
Result in April 2026: +47% organic sessions on the restructured guides. 91% of them are now cited in at least one AIO.
Traffic doesn’t come back on its own. It comes back because we earned the citation.
Paid Search Holds Steady: +10% CTR When an AIO Appears
One point the Seer Interactive study documents—and that few public analyses address: paid CTR behavior facing AI Overviews.
The numbers:
- When an AI Overview displays: Average paid CTR of 16.2%
- When no AIO displays: Average paid CTR of 21.8%
First reading: yes, the AIO drops paid CTR by -26%. But this drop is far gentler than organic CTR decline (-73% between a SERP without AIO and a SERP with AIO for an uncited page).
Second reading, counterintuitive: between early 2025 and February 2026, paid CTR on SERPs with AIO has slightly increased, rising from 14.6% to 16.2%. That’s +11%.
Why?
Because users who click despite an AIO’s presence are qualified differently. They’re not seeking a quick answer. They’re seeking a purchase, an advanced comparison, a spec sheet, a stock check.
The AIO filters pure informational clicks. It lets transactional clicks through.
Result: paid search captures a more qualified slice of residual traffic.
I observe it with a maritime equipment client. Google Ads budget of €4,200/month. Since AIOs rolled out on their « how to choose children’s life jacket », « what wetsuit size for neoprene » queries, their Shopping ad CTR has climbed +14% between June 2025 and March 2026.
Why? Because people who clicked just « to see the answer » don’t click anymore. They read it in the AIO. Those who click now want to buy.
CPC hasn’t exploded (+8% only). Conversion rate rose +11%. ROAS went from 4.2 to 5.1.
Google didn’t kill paid. It refined it.
CTR Decline ≠ Traffic Decline: The Impressions Trap
One point the Seer Interactive study stresses—and that I see misunderstood in 80% of clients calling me about « AIO traffic loss »:
A CTR decline doesn’t always mean an absolute click decline.
Why?
Because your impressions can rise at the same time.
Typical example: you’re cited in 15 AI Overviews in January 2026. In March 2026, you’re cited in 38 AIOs. Your impressions double. Your average CTR on those queries drops from 3.2% to 2.1% (because of AIO).
Net result: you gain +31% of clicks despite a -34% CTR decline.
This is what Seer calls in its study « impression growth masking CTR decline. »
I saw it live with a professional training client. 240 pages of certification guides. Between August 2025 and February 2026:
- Average CTR: -28%
- Impressions: +140%
- Organic clicks: +67%
How?
Because Google started displaying AIOs on 900 new long-tail queries tied to their content. Queries they never appeared on before (positions 18-30, invisible).
Now they appear in AIOs. With low CTR (1.8%), but gigantic impression volume.
Bottom line: +67% organic sessions in six months. Despite a « CTR decline. »
When a client tells me « I lost CTR because of AIOs », I look at three metrics in this order:
- Absolute click volume (not CTR)
- Impression volume (is it rising or falling?)
- Citation rate in AIOs (how many of my pages get referenced?)
CTR alone says nothing. You need context.
How to Build Content Google Cites in AIOs
The question everyone asks me: what makes a page cited in an AI Overview?
I can’t give you an algorithm. Google doesn’t publish it. But I can give you what I observe across 200+ pages my clients track that are consistently cited in AIOs over 8 months.
1. Primary data beats reformulation
Cited pages contain at least one element of data Google can’t generate itself:
- Comparison table with measurements (duration, weight, price, dimensions)
- Test result (« we tested 12 models over 6 months »)
- Numbered case study (« 34% fewer returns after process shift »)
- Annotated graph, technical diagram, dimensioned plan
Google can reformulate a definition. It can’t invent a 15-row table with data only you’ve measured.
2. Clean markdown structure
Cited pages have crisp HTML:
- Consistent H2 > H3 hierarchy
<ul>/<ol>lists (not bulleted paragraphs)<table>tags with<thead>and<tbody>- No 400-word text blocks without breaks
Google parses. If it can’t understand the structure, it won’t cite.
3. Direct answer in the first 150 words
Cited pages answer the question within the opening 150 words. Not after 3 intro paragraphs.
Example: query « what insulation thickness for unfinished attics RT 2012 ».
Page A (not cited): « Attic insulation is a major energy renovation issue. Since RT 2012 was implemented… [300 words]. The answer: 300 mm minimum. »
Page B (cited): « RT 2012 requires 300 mm minimum insulation in unfinished attics (R ≥ 7 m².K/W). For renovations, the typical range is 320-350 mm. Here’s the breakdown by climate zone: [table]. »
Google cites B. Not A.
4. Freshness or authority
Cited pages are either:
- Recent (updated <6 months)
- Or from a domain with strong topical authority (consistent backlinks, dense semantic cluster)
An isolated, never-updated page with no internal linking has 90% odds of being never cited. Even if the content is solid.
5. No filler
Cited pages don’t contain empty phrases like « in today’s digital landscape », « it should be noted that », « since time immemorial ».
Google has trained on trillions of tokens. It spots padding. It doesn’t cite it.
If your intro runs 4 paragraphs without a single number, you won’t be cited.
What This Means for Your E-Commerce Site in 2026
Let’s summarize.
AIO CTR is climbing. But it remains below a non-AIO SERP. And it’s sharply unequal between cited pages (+2.1%) and non-cited pages (0.9%).
For an e-commerce site, that means three things:
1. Your product cards and catégories aren’t at risk
AIOs show up on only 5% of transactional queries. Google wants to sell you ads, not answer « buy X » with generated text.
Your catalog stays protected. Keep optimizing it. Keep building category clusters. Nothing changes.
2. Your informational content has to earn its citation
Your buying guides, « how to choose » pages, FAQs — they’re all in AIO crosshairs. 86% of question queries display an AIO. 95% of comparison queries do too.
If your guide lacks proprietary data, tables, tests, sourced numbers — it will sink.
You need to shift from « generic reformulated guide » to « guide bringing data Google can’t create ».
3. Paid search gets sharper on certain queries
AIOs filter pure informational clicks. They let qualified clicks through. Result: your paid CTR drops in volume but rises in quality.
On queries where an AIO appears, test a bid increase. CPC climbs slower than conversion rate. You gain ROAS.
I’ve done this with 8 clients since November 2025. 7 out of 8 saw ROAS jump +12% to +34%.
A complete example
Camping gear client. 1,200 SKUs. €6,800/month SEA budget. In October 2025, Google rolls out AIOs on 80% of their guide queries (« how to choose cold-weather sleeping bag », « synthetic vs down differences », etc.).
Between October 2025 and January 2026: -31% organic traffic on guides. No dip on the product catalog.
We restructure 22 guides. We add:
- Comparison tables (comfort vs limit temperature, weight, price per gram)
- Annotated technical diagrams (down compartment types)
- Field tests (« tested 14 nights between -8°C and -15°C »)
We republish between February and March 2026. April 2026 result:
- 19 of 22 guides cited in at least one AIO
- +58% organic sessions on restructured guides
- Average CTR: 2.3% (vs 3.1% pre-AIO, vs 0.8% pre-restructure)
Paid side: we raise bids +18% on « how to choose » queries (AIOs present). CPC rises +9%. Conversion rate climbs +22%. ROAS shifts from 4.7 to 5.9.
Total March 2026 vs September 2025: +12% organic + paid revenue on stable ad spend.
Google doesn’t kill SEO. It kills lazy SEO.
Losing traffic since AIOs arrived?
First call: live audit of your visibility in AI Overviews. We check your pages, spot ones that can get cited, restructure what needs it. You walk away with a numbered action plan.
Book a strategic call — 45 minFrequently Asked Questions
Is AI Overviews CTR really bouncing back?
Yes. Per Seer Interactive’s study (2.43 billion impressions), average CTR jumped from 1.3% in December 2025 to 2.4% in February 2026—a +85% leap in two months. But it stays below a non-AIO SERP’s 3.8%.
What’s the CTR gap between cited and non-cited pages in an AIO?
A cited page in the AI Overview gets 2.1% average CTR. A page on the same SERP but not cited drops to 0.9%. That’s +133% in favor of cited pages.
Do AI Overviews appear on all query types?
No. They show on 86% of question queries, 95% of comparisons, 36% of informational, but only 5% of transactional queries (per Seer Interactive).
Is paid search hit by AI Overviews?
Yes, but less hard than organic. When an AIO appears, paid CTR shifts from 21.8% to 16.2%. Yet between 2025 and 2026, it climbed +11%—a sign clicks are getting more qualified.
How do I get cited in an AI Overview?
Bring primary data (tables, tests, measurements), structure clean HTML (lists, H2/H3, tables), answer directly in the first 150 words, maintain content freshness, and skip filler language.
