Publishers have spent two years describing the same shape: rising impressions, shrinking clicks, the pattern known as the crocodile effect. New data on Google Shopping ads shows the opposite trend line, and the divergence points to a deliberate revenue mechanism rather than a coincidence.
Mike Ryan of ecommerce advertising firm Smarter Ecommerce analyzed roughly 175 billion ad impressions and found that median Shopping ad impressions fell from about 1.85 million to 1.4 million between mid-2025 and mid-2026. Over the same period, median click-through rate rose from 1.20 percent to nearly 1.55 percent. Ryan published the findings on LinkedIn, framing the pattern against the publisher experience directly: “Here’s an odd one. Ever since AIOs launched, publishers have been gaining impressions and losing clicks: the dreaded SEO crocodile effect. For Shopping ads, it’s the opposite. But why?”
Ryan’s explanation centers on query-level probability scoring. His theory holds that Google evaluates the predicted click-through rate of a given search before deciding whether to show an AI Overview, and preferentially deploys the AI answer on queries least likely to convert into a paid click. That selection process removes low-intent searches from the ad auction pool entirely, shrinking the denominator of total impressions while leaving a query mix skewed toward users more likely to click.
The mechanism differs sharply from what has driven publisher losses. Publishers lose clicks because AI Overviews answer the query directly inside the search results page, satisfying informational intent without a visit. Shopping ads, by Ryan’s account, lose impressions instead: Google is not answering the transactional query so much as withholding it from the standard ad slot, which then makes the remaining ad impressions convert at a higher rate. One pattern is about traffic diverted. The other is about inventory selectively pruned.
Ryan calls this a bridging strategy rather than a stable end state. He expects Google to shift the current either-or split, in which a query gets either an AI Overview or a Shopping ad slot, toward a both-and model as AI-native ad formats mature and Shopping ads begin appearing inside AI Overviews themselves at scale. Search Engine Roundtable, which first reported the data, noted that Ryan’s post did not include a breakdown of raw click or revenue totals alongside the median percentages, so the underlying volume shift beneath the CTR gain remains unverified outside his aggregate figures.
For search and ecommerce teams reading their own Google Ads dashboards, the practical read is that a falling impression count on Shopping campaigns is not automatically a budget or bid problem. If Google is systematically routing lower-probability queries away from the Shopping auction and into AI Overviews, campaign-level CTR gains can mask a shrinking addressable query set rather than reflect improved targeting. Retail marketers should segment impression volume by query category over the next quarter to see whether the decline concentrates in informational or comparison-shopping terms, since that is where an AI Overview substitution is most likely, and should watch for early signals of Shopping ads embedded directly inside AI Overviews before locking in next year’s paid search budget assumptions.
Search Engine Roundtable, reporting by Barry Schwartz, published September 7, 2026.
