Organic CTR Fell 61 Percent on AI Overview Queries in One Study. Revenue Grew in Another. Track Citations and Revenue, Not Just Clicks.

Measured published an incrementality analysis across its dataset of 139 brands and reported that median incremental revenue and orders both grew after Google's AI Overviews rolled out broadly in September 2025, while spend remained flat. Brands that also measure retail in-store revenue saw retail iROAS climb from $1.35 to $1.47 in Q1 2026. Separately, Seer Interactive's study of 3,119 informational queries across 42 client organizations found organic CTR on queries with an AI Overview fell 61%, from 1.76% in June 2024 to 0.61% in September 2025. These are different datasets measuring different things, so neither proves the other. Together they say the same thing: click-through rate alone no longer tells you whether search is working.

The CTR Panic Measured One Proxy

The "AI is killing organic" story rests on real data that is incomplete. Seer's figures are for informational and educational queries, the ones most exposed to AI Overviews, across its own clients, and Seer reports paid CTR on those queries down 68% as well. Seer notes that this challenges its own initial assumptions about causation: queries that later showed an AI Overview already had lower organic CTR in June 2024 (1.76% against 2.74% for queries without one), and organic CTR on queries with no AI Overview also fell (46.2% year over year in its Q3 2025 comparison). Pew Research Center's analysis of 900 U.S. adults' browsing found users who encountered an AI summary clicked a traditional result link in 8% of visits, against 15% for users who did not.

CTR counts clicks. Revenue counts money. Measured's dataset looked at business outcomes and found median incremental revenue and orders grew while spend stayed flat. Measured does not say AI Overviews caused the growth, and its 139 brands are its own measurement clients, not the market. What the data does show is that a falling click rate and rising incremental revenue can coexist, so a team reading only CTR can reach the wrong conclusion.

What the Evidence Says About AI Overviews and Citations

Citation is a separate signal from the click. In Seer's data, brands cited inside an AI Overview earned 35% more organic clicks and 91% more paid clicks than when they were not cited. Seer states the caveat itself: it cannot prove citation causes the higher CTR, because brands with stronger authority may simply be more likely to be cited. Read it as an association worth tracking, not a lever with a guaranteed return.

AI-referred visitors can be worth more per visit, in at least one cohort. Attrifast's 200-site study, a cohort of Stripe-connected SMB sites that Attrifast says is not an industry-wide truth, found AI-engine traffic to B2B SaaS converting to a Stripe payment at 2.7% versus 1.4% for Google organic on the same landing pages. In that cohort, revenue per visitor was $1.42 from Perplexity, $1.18 from Claude and $0.87 from ChatGPT, against $0.61 for Google organic. The cohort is one vendor's customers, so treat the numbers as a direction to test on your own data.

Your analytics may be hiding some of it. Attrifast's server-side fingerprinting reclassified a median 34% of GA4 "Direct" traffic as AI-referred in that cohort (interquartile range 21% to 47%). That is a modeled estimate, not an observed count, and the share on your site will differ. The point stands: sessions arriving without a usable referrer land in Direct, so a growing Direct bucket is not automatically a brand-awareness win.

Revenue Quality in the Same Cohort

In the same Attrifast cohort, first-month SaaS customer value was $44.10 for ChatGPT-sourced customers versus $28.70 from Google organic, about 54% higher (Claude $57.40, Perplexity $51.20, but Gemini $23.40 and AI Overviews $21.90), and blended 30-day churn across AI engines was 9.2% for SaaS customers versus 14.4% for Google organic. These are one vendor's cohort figures. They justify measuring customer value by source on your own Stripe data; they do not establish an industry rate.

What This Means for How You Measure

Do not run a marketing budget off CTR alone. Add the measures CTR cannot see:

Track citation presence alongside clicks. Record whether your brand appears in the AI-generated answer for the queries that matter to your pipeline, and watch it next to organic CTR and revenue.

Track revenue per AI-referred visitor on your own data. Count sessions with a recognized AI hostname or UTM, then reconcile them with your payment records. Do not borrow another site's benchmark as your own.

Treat last-touch attribution as incomplete. Sessions with no referrer are sorted into Direct before any channel rule sees them. Server-side referrer parsing and incrementality tests such as Measured's give a fuller read than a GA4 default channel report.

The discipline of Machine Relations is built around this shift: earning the citations AI engines extract, measuring whether those citations drive discovery, and compounding that presence over time.

FAQ

Did AI Overviews actually increase revenue for brands?

Not on this evidence alone. Measured's analysis of its 139-brand dataset found median incremental revenue and orders grew after the AI Overviews rollout while spend remained flat, and brands with retail in-store measurement saw iROAS rise from $1.35 to $1.47 in Q1 2026. That is an association inside one vendor's dataset, not proof that AI Overviews caused the growth.

Why can CTR fall 61% while revenue holds up?

CTR measures clicks, not business outcomes. Seer Interactive measured the 61% organic CTR drop on informational queries with AI Overviews across its own clients. Measured's dataset is a different, broader measurement of outcomes. The two do not contradict each other, because they count different things for different brands.

How much AI search traffic is my analytics missing?

It depends on your site. Attrifast estimates that a median 34% of GA4 "Direct" traffic in its 200-site cohort is AI-referred, an estimate produced by server-side fingerprinting rather than an observed count. Server-side referrer parsing can recover some of these sessions, but standard GA4 configurations cannot separate them from direct visits.