Click-through rate sounds like the simplest performance number in digital marketing: divide clicks by impressions, multiply by 100. But in 2026, a 5% CTR is not the same 5% everywhere. Google Ads, Meta, Search Console, programmatic platforms and ChatGPT all count different events, and AI Overviews are quietly moving the denominator.
From novelty to pricing lever
CTR became a metric because clicks were the first web behaviour that could be measured. HotWired’s 1994 AT&T banner reportedly generated a 44% click rate. By 2009, comScore and Starcom found only 16% of US internet users clicked a display ad in a month, and 8% of users produced 85% of clicks.
As comScore’s Linda Anderson said at the time, click-only optimization assigns “no value to the 84 percent of Internet users who don’t click on an ad.”
That decline also explains why search engines used CTR to price ads. Google’s AdWords Select launched in February 2002 with cost-per-click pricing and began weighting bids by historical CTR. An advertiser bidding $1.00 with a 1% CTR is worth $10 per thousand impressions; one bidding $0.50 with a 5% CTR is worth $25.
Google later folded this into Quality Score. Expected CTR is a position-neutral prediction, adjusted for ad position and assets, and recalculated at auction time using search terms, device and other signals. It is not the raw CTR column in your report.
Why your CTR does not match your analytics
The IAB and MRC guidelines describe four click stages: initiated, measured, received and resolved. A platform can count a measured click when it sends a redirect; an advertiser may only record a received click when the request reaches its server. Latency, abandoned loads and broken pages all create a gap.
The denominator also changes by platform. Google Ads divides by impressions. Search Console only counts impressions when a result appears on a page the user actually viewed. Meta separates a CTR for all interactions from a link click-through rate for destination clicks. And AdSense changed Multiplex counting on June 23, 2026, so ad request CTR no longer exceeds 100% per slot.
AI Overviews are rewriting organic CTR
For organic search, CTR is now the clearest signal of what generative answers take away. SISTRIX found position-one CTR in Germany falls from 27% to 11% when an AI Overview appears. Seer Interactive measured organic CTR down 61% and paid CTR down 68% on informational queries carrying the feature. A randomized study estimated a 39.8% outbound organic click loss.
Not every CTR shift is caused by your campaign. Mike Ryan of Smarter Ecommerce analysed 175 billion impressions and found median Google Shopping impressions fell from roughly 1.85 million to 1.4 million while median CTR rose from 1.20% to nearly 1.55%. The likely explanation: Google is removing low-probability impressions via AI Overviews, making CTR look healthier without any advertiser change.
A working framework for performance teams
Treat CTR as a diagnostic, not a KPI. A click is attention, not an outcome. Use it at three levels:
- Relevance: a low CTR on high-intent search ads often flags weak ad relevance or poor expected CTR.
- Distribution: falling organic CTR may mean AI Overviews are absorbing the query, not that your SEO has slipped.
- Integrity: sudden CTR spikes can signal bots, accidental clicks or a platform-side denominator change.
Also pair raw CTR with conversion rate and view-through measurement for display and social. Lunio found invalid rates in retail search accounts rose from 4.18% in Q4 2025 to 5.54% in Q2 2026, and DoubleVerify reported AI bots can generate up to ten times more clicks than humans in unprotected campaigns.
In ChatGPT ads, early Similarweb data put ad CTR at 0.50% to 0.68% overall, while Criteo self-reported roughly three times comparable formats. The definition behind those numbers matters before you optimize anything.
Source: PPC Land
