If your Shopping or PMax campaigns have been looking more efficient lately — CTR up, maybe ROAS holding steady — your instinct might be to leave well enough alone. That instinct might cost you Q4.
Search Engine Land reported September 8 on benchmark data covering approximately 175 billion impressions across hundreds of advertiser accounts: median Shopping ad impressions per account fell from roughly 1.85 million in mid-2025 to around 1.4 million in mid-2026, while CTR rose roughly 20% over the same period. Analysts are calling this the "reverse crocodile" effect. The two lines on the chart diverge in opposite directions: impressions down, CTR up. And at first glance, it looks like progress.
It isn't necessarily.
What the data actually says
A CTR lift sounds like better ad relevance or better creative. And it might partly be that. But according to Search Engine Land's coverage of the benchmark, click numbers were "relatively flat or slightly down" across most accounts despite the gain. The math plays out the same regardless: if impressions fall faster than clicks, CTR rises while total reach shrinks. You're still losing ground on volume.
The efficiency metric improved because the denominator shrank — not because the numerator grew.
Why AI Overviews is the likely explanation
The leading hypothesis, which Search Engine Land has described as plausible but unproven, is that Google AI Overviews is absorbing the top of the shopping funnel. When a user asks "best running shoes under $100," AI Overviews gives them a curated answer with product suggestions. Many of those users never scroll down to the Shopping ad carousel. So the impression never happens.
What remains in the Shopping ad auction — at least the hypothesis goes — are users with stronger purchase intent, people already past the discovery phase. They click more often per impression. Hence: CTR up, impressions down, clicks flat.
According to ALM Corp's analysis, AI Overviews now appear on roughly 14% of Shopping-related queries in the US, up from near-zero 18 months ago. That's a material share of funnel to have removed from your reach without any account-level visibility into it.
Why this matters more than your efficiency metrics this month
Here's the problem with leaning into this apparent efficiency gain: you can't see what you're missing.
Your CTR is measured against the impressions you received. If Google withheld 400,000 impressions per month from your account — impressions that might have converted at a lower rate, but converted — your standard Shopping reports don't show you that gap. They just show a cleaner CTR.
This creates a real Q4 risk. If your account has quietly been deprived of upper-funnel impression volume across the past year, your demand pool may be smaller going into Black Friday than it was in 2025. The brands that built awareness through that upper-funnel exposure, whether through Shopping or video inventory, will have a warmer audience to convert. You might not.
There's also a budget signal buried in the data. If Google's auction is serving fewer impressions, your campaigns may be burning through budget faster per impression at the bottom of the funnel — competitive, high-intent queries — without filling the pipeline from discovery queries the way they used to.
What to check in your account
Your account won't have an "AI Overviews" filter in the Search Terms report. You can't see which queries Google served to AI Overviews instead of your Shopping ads. But you can see the shape of what's happening:
- Pull your Shopping impression volume trend on a monthly basis going back 12 months. Is it flat or declining?
- Compare that to your click trend. If impressions are down more than clicks, your CTR math is doing what the benchmarks describe.
- Look at your reach metrics if you're running PMax — does the "Audience reach" signal support an impression volume story?
- Check your Google Analytics organic shopping traffic for the same period. If discovery-stage organic traffic from Shopping is also flat or declining, you may have a broader AI Overviews exposure issue.
None of this tells you definitively what to do. The underlying benchmark data is observational — a 12-month correlation, not a controlled experiment. But the pattern Search Engine Land reported across 175 billion impressions is hard to dismiss.
The more useful question isn't "what happened to my CTR?" It's "what happened to my total clicks, and where did the top of my funnel go?"
If you're not sure how healthy your Shopping campaigns are heading into Q4, a free account audit surfaces the signals that don't show up in your standard reporting. Run yours at gromerce.com/audit.
Your CTR going up while total clicks go sideways isn't a performance improvement — it's a measurement artifact that hides a shrinking audience.
Sources: Search Engine Land, ALM Corp, September 2026
