Blog
The AI Overviews Paid Search Stat Everyone Quotes Is a Year Out of Date
By Jared Schroder, CEO and Founder, SEMoptimize
Somebody put a slide in front of me this spring saying AI Overviews had cut paid search click through rate by about two thirds, from 19.70% to 6.34%, and I nodded along, because I had used the same figure in a client note a few months earlier. The number is real and it traces to a named study with a stated sample and a stated date. What I had not done, and what almost nobody passing it around has done, is check whether the firm behind it had published anything since. It had. In April 2026 the same team, on a paid sample roughly 270 times larger, put paid CTR on AI Overview queries at 16.21%, and both studies are still up with neither retracted. So this is a trace of one statistic, what its authors published six months later, and what the new inventory actually costs, which is a question about targeting control rather than clicks.
Where does the 19.7 percent to 6.3 percent paid CTR figure come from?
It comes from Seer Interactive, published November 4, 2025, in "AIO Impact on Google CTR: September 2025 Update" by Tracy McDonald, Product Development Lead in Seer's R&D group. Under a heading reading "Paid CTR for AIO Queries" the post lists June 2024 at 19.70%, July 2025 at 3.26%, September 2025 at 6.34%, and an overall decline of 68%.
This is genuinely a paid number, which matters because the commonest error in this topic is quoting the 61% organic collapse from the same study as though it were the paid one. Paid CTR here is ad clicks divided by ad impressions out of Google Ads, measured separately from organic CTR out of Search Console, on queries carrying an AI Overview.
The stated sample is 3,119 search terms across 42 client organizations, June 2024 through September 2025, with 25.1M organic impressions against 1.1M paid impressions, and that gap between the two counts is why the paid half of the study reads as it does.
Why that number should not be in your 2026 deck
Seer's scope note disqualifies most of the uses the figure gets put to: "This analysis covers only informational/educational queries. Your results may differ if you operate in different query types (commercial, transactional, local)." A study of informational queries became a claim about paid search generally, and if your account lives on product and purchase language the 6.34% was never measuring you.
Then there is the 1.1M paid impression count, which Seer flags rather than leaving a reader to find: "The paid impression sample (1.1M) is significantly smaller than organic (25.1M)... This represents a limitation for the paid analysis. Findings are directionally consistent but have wider confidence intervals than organic." The same section reports standard deviations of "3-5 percentage points for paid," and put a 3 to 5 point band around a 6.34% mean and the difference between a crisis and a soft quarter sits inside the error.
The series also moves far enough that the month you quote does much of the work in your argument. Paid CTR on AI Overview queries runs around 11% before July 2025, drops to 3.26% that month, and is back to 6.34% by September. Seer calls that drop "striking" and confirms no cause for it, so I will not hand you one either.
The post also disagrees with itself twice, which is only visible because Seer publishes enough detail to check: "Overall decline: 68%" against "Paid CTRs crashed 65%" in its own summary, and cited paid CTR at 7.89% in one passage and 11.05% in another. Figures that unsettled inside their own post were never meant to be quoted to two decimals, which is the failure mode I have written about in the context of over reliance on Google Ads automation.
What the same researchers found six months later
In April 2026 the same team put paid CTR on AI Overview queries at 16.21%. The study is "AIO Impact on Google CTR: 2026 Update", published 2026-04-24 by McDonald with Hannah Cooley and Marketa Williams, on 53 accounts, 5,471,127 distinct queries and 296.9M paid impressions, January 2025 through February 2026. Roughly 270 times the paid impression volume of the earlier study is why it is the one I plan against.
| Month | Paid CTR, AIO shown | Paid CTR, no AIO |
|---|---|---|
| Jan 2025 | 14.64% | 25.98% |
| Jun 2025 | 14.17% | 18.06% |
| Sep 2025 | 15.48% | 19.14% |
| Dec 2025 | 15.14% | 20.96% |
| Feb 2026 | 16.21% | 21.85% |
Look at September 2025 there. It reads 15.48% paid CTR on AI Overview queries against 6.34% for the same month, same construct, same firm. They disagree by 2.4x because they are not one cohort measured twice: the November 2025 study covered informational and educational queries across 42 organizations, and the 2026 study spans informational, commercial and transactional intent across 53 accounts. Stringing 19.7 to 6.3 to 16.2 into a trend line would be a chart of nothing.
Seer answers that objection with an informational only table, the segment the earlier study measured, where citation means the domain is linked inside the AI Overview rather than merely facing one.
| AI Overview status, informational queries, 2025 full year average | Average organic CTR | Average paid CTR |
|---|---|---|
| Brand cited | 2.07% | 15.74% |
| Brand not cited | 0.94% | 11.19% |
| No AIO | 3.35% | 19.75% |
Restrict to informational and the paid floor is 11.19%, so the 6.34% does not reproduce at scale, and the accurate word for it is superseded rather than wrong. Seer's own reading is the line I would put on the wall: "Whatever AIO is doing to organic results, it is not doing the same thing to paid ads. Paid and organic are operating in different environments on the same SERP, and you should plan accordingly."
Search Engine Land ran the first study as "Google AI Overviews drive 61% drop in organic CTR, 68% in paid", which is where most of the circulation comes from, and Danny Goodwin covered the revision the day it published, reporting that "Paid search is far more stable." The correction was reported and did not travel.
The paid CTR that is actually falling has no AI Overview on it
Paid CTR on queries with no AI Overview fell from 25.98% in January 2025 to 21.85% in February 2026, about 16% down, while paid CTR on queries that did show one went up over the same window. That inversion changed how I read the topic, because the advice built on the panic number is to move budget toward inventory that does not trigger AI Overviews, and that is the only paid segment in this data going the wrong way. The November 2025 study points the same way, with paid CTR on non AIO queries down from 19.1% to 13.04%, and two cohorts agreeing on a direction is weaker than one controlled measurement while still being the same direction twice.
Nobody has established why, and Seer will not try: "We cannot claim causation." Their data shows queries prone to triggering AI Overviews had lower CTR before AI Overviews appeared on them, so presence and CTR are entangled from the start, and they print an organic pattern they cannot explain, that "AIO is not the only force compressing organic CTR."
Which query classes actually carry AI Overview exposure?
Informational queries show an AI Overview about 36% of the time, commercial queries about 8%, and transactional queries about 5%, across the 49,358 queries Seer tracked for the 2026 study. Seer gives the mechanism: "When someone searches for a local plumber or a specific product, the Shopping Box or Maps result is the answer. AIO isn't competing with that."
| Query type | AIO rate | Organic impressions in sample |
|---|---|---|
| Comparison (X vs Y) | 95.4% | 79,436 |
| Review queries | 86.3% | 21,221 |
| Question (what/why/how/is/are) | 85.9% | 1,408,843 |
| Price / Cost / Buy | 83.4% | 192,911 |
| Best Of | 81.3% | 105,279 |
| Near Me | 76.9% | 188,606 |
The 95.4% at the top is the row everybody will screenshot, and it rests on 280 queries, so I read it as a strong hint rather than a rate. Question format at 85.9% is what I plan against, since it sits on 1,413 queries and 1.4M impressions and will not move much on the next pass.
Query length is where I had to drop an assumption I had been carrying. Single word informational queries trigger an AI Overview 27.3% of the time, two word queries 34.8%, three word 31.1%, and queries of four words or more 35.5%, so longer queries trigger them slightly more often rather than less. Seer says it plainly: "Don't assume short queries are safe... The idea that broad, short queries are less AIO-affected is not supported by this study." The segmentation that survives contact with this data is intent, and a long tail informational question is among the most exposed shapes on the board rather than a hiding place.
For one group of advertisers the question is moot: Google's documentation says it does not show ads in AI Overviews for "sensitive verticals like adult, alcohol, gambling, finance, healthcare, politics, and more," so for those sellers the placement does not exist today.
What AI Overviews actually cost a paid search program is query control
Google conditions eligibility for ads inside AI Overviews on giving up query level targeting, in its own words: "This is why Google requires AI-powered targeting solutions like broad match on Search or the keywordless targeting technology available through AI Max for Search campaigns, Performance Max campaigns, Shopping campaigns, or Dynamic Search Ads, to match relevant ads." Keywordless targeting means matching without advertiser supplied keywords. The same help page says you cannot target the placement, cannot opt out of it, that ads serving there report as Top Ads, and on reporting it is unambiguous: "Google Ads currently doesn't offer segmented reporting when ads show within Search AI Overviews."
The case against a relaxed reading of the CTR data belongs here first, because parts of it hold. AI Overviews are moving down funnel, which is what Semrush's intent mix shift from 91.3% informational in January 2025 to 57.1% by October 2025, across more than 10 million tracked keywords, describes, and Seer's published bet is that transactional prevalence "stays below 10% through Q3 2026," an admission it might not, so today's 5% is a snapshot rather than a floor. The upturn in paid CTR is two months at the end of the series and Seer will not lean on it either: "We aren't suggesting you forecast growth or recovery based on only Jan-Feb 2026 data." Their model, built on twelve months of 2025 data, predicted continued decline and was wrong, and one account carries 47% of the impressions in their informational not cited segment. Every figure here is a cohort average with a 3 to 5 point standard deviation on paid, so an account concentrated in comparison and question format queries lives in the 85.9% bucket rather than the 36% average.
Grant all of it and the control problem is still standing, because it holds under either CTR reading. The terms are that you adopt broad match or keywordless targeting to be eligible, take a placement you did not choose and cannot decline, and get no data telling you when you were in it. For a shop that builds accounts out of query data, which is how we approach query level paid search optimization, that is the harder problem, because a CTR figure you can eventually measure yourself and this you cannot measure at all.
The organic side sharpens it. On 2026-06-03 Google launched Search Generative AI performance reports in Search Console, covering impressions in AI Overviews and in AI Mode, the conversational surface Sundar Pichai says has merged with AI Overviews into "one seamless Search experience" and passed a billion monthly active users. That report is thin, with impressions, pages, countries, devices and dates but no clicks, no CTR and no query data, starting 2026-05-18 and no backfill. Thin as it is, an SEO can now see per URL when their page appeared in an AI Overview and an advertiser still cannot see when their ad did, so the channel billing by the click has less visibility into the new surface than the channel that bills nothing. More ad formats were announced on 2026-05-20, framed by Google as testing rather than availability, which means the environment the 2026 data describes will be gone by the time anyone measures it again.
What to check before your next budget conversation
Four things, none of which cost anything.
Check the publication date and the paid impression count on any AI Overview CTR statistic before it goes in a plan. Those two facts would have caught the 6.34% for me, and both took ten minutes to find once I looked.
Make every CTR number in the deck carry the word paid or organic in the same sentence, because the 61% organic decline and the 68% paid decline came out of one post on one day and most of the damage here is one being carried over as the other.
Segment your own query report by intent rather than by length, and see what share of your impressions sit in question and comparison shapes, because exposure concentrates by query shape and your mix will not match a 53 account cohort.
Find what share of your spend already runs through broad match, AI Max, Performance Max, Shopping or Dynamic Search Ads, since that is the part of the account already eligible for a placement you cannot see in reporting. If you sell in one of the sensitive verticals Google names, start there, because the answer may be that none of it applies yet.
Frequently asked questions
Did AI Overviews really cut paid search CTR by two thirds? Not as a current statement about paid search. The 68% decline comes from Seer Interactive's November 4, 2025 study of 3,119 informational search terms across 42 organizations and 1.1M paid impressions, and in April 2026 the same team, on 5,471,127 queries and 296.9M paid impressions, put paid CTR on AI Overview queries at 16.21% in February 2026.
Is paid CTR or organic CTR more affected by AI Overviews? Organic, by a wide margin, in the data that exists. Seer's 2026 study reports organic CTR on AI Overview queries bottoming at 1.3% in December 2025 and recovering to 2.4% by February 2026, while paid CTR on those queries moved from 14.64% to 16.21%. Seer states that they cannot claim causation, so none of that establishes AI Overviews as the cause.
Which searches are least likely to show an AI Overview? Transactional ones, at about 5% prevalence against 36% for informational, across the 49,358 queries Seer tracked for the 2026 study. Query length will not help, since single word queries trigger an AI Overview 27.3% of the time and queries of four words or more 35.5%. Low exposure is also not safety, because paid CTR on queries with no AI Overview is the segment that fell, 25.98% to 21.85%.
Can I see in Google Ads when my ad served inside an AI Overview? No. Google Ads Help states that "Google Ads currently doesn't offer segmented reporting when ads show within Search AI Overviews," that ads serving there report as Top Ads, and that advertisers can neither target the placement nor opt out of it. Search Console gained a report for organic pages on 2026-06-03 with impressions but no clicks, so what visibility exists favours the unpaid channel.
