Cyrus Shepard estimates that appearances inside Google’s AI Overviews and AI Mode earn a click-through rate of about 0.9 percent, and that blocking a site from those features also cost it clicks in standard web results. Shepard published the figures as a small case study on LinkedIn, and Search Engine Roundtable reported them on Friday, October 30. The result matters because it challenges the assumption that AI citations and organic listings compete for the same click.
Shepard used Google’s switches for declining generative features to hold one site out of AI Overviews and AI Mode for a test window, then restored its eligibility. During the window, the site’s AI impressions dropped to zero immediately, he said. The AI answers themselves did not vanish. Shepard told Search Engine Roundtable that AI Overviews appeared on the relevant results throughout the test. His site was cited in them at the start and again at the end, but absent for the middle stretch.
The 0.9 percent figure is not a direct measurement. Shepard derived it by taking the site’s overall click-through rate and impressions, then subtracting the AI impressions he could identify during the experiment period. He described it as an estimate. He also put AI impressions at about 8 percent of the site’s total impressions, so in this dataset the AI surfaces were a small slice of overall visibility.
The more surprising result concerns the blue links. Regular organic clicks also fell when the AI features were disabled, and overall impressions dropped as well. Sitewide average position actually improved, which Shepard attributed to two or so fresh posts plus brand search. The pages that had earned the heaviest AI exposure told a different story: performance dipped during the block and recovered once visibility returned.
Shepard reads this as a “Search SERP Halo Effect”: each extra spot where a brand shows on a results page raises the odds that a searcher clicks something from it. Greg Sterling of Near Media offered supporting context in the comments. His firm’s research found that AI Overviews can create brand awareness that shows up in later searches and clicks, though not consistently.
The limits are large, and the source states several of them. This is one site. Shepard calls the study small, and the write-up gives no query count, no test duration, and no sample size. Barry Schwartz of Search Engine Roundtable also flagged that AI Overviews stayed on the page throughout, so the test compares being cited with not being cited. It does not compare results pages with and without an AI answer.
Two confounds sit inside the reported data. Fresh posts and brand search moved average position during the period, and neither is separated from the AI toggle. The write-up also does not say whether the lost organic clicks came from fewer impressions, a lower click rate, or both. A drop in impressions alone would reduce clicks without any halo.
Even so, the case study fills a gap. Schwartz wrote that he had not seen anyone else publish opt-out data like this, probably because few site owners want to risk the visibility. Until a larger test appears, the 0.9 percent number is a single estimate from a single property, and the halo reading is a hypothesis.
Teams weighing the AI opt-out should log branded and non-branded organic clicks separately before touching the controls, because lost clicks in classic results are the cost this case study suggests they could be paying.
Cyrus Shepard’s LinkedIn case study, reported by Search Engine Roundtable (Barry Schwartz) on October 30, 2026.
