Site lost 21% clicks after Google AI opt-out test

Cyrus Shepard from Zyppy SEO took his site out of Google's AI features for four weeks to test the real click impact. He published the results and it got covered on Search Engine Journal.

SEJ on Google's AI click impact

During the opt-out the site vanished from AI Overviews and AI Mode. Impressions dropped to zero almost instantly. When he flipped the setting back on after 28 days the impressions came back nearly as fast on the same searches.

He compared clicks on identical queries and positions to calculate a CTR of about 0.9% for the AI impressions. That's double his site's normal CTR of 0.4%. With AI making up roughly 10% of impressions he estimates they drove 21% of all search clicks. Adding a small lift to regular organic results it hits 23%.

Shepard calls it a halo effect. Seeing the site multiple times on the page apparently makes people more likely to click one of the regular listings. He says treat the numbers as a directional yardstick for his single site and not assume they apply everywhere.

Still he is keeping the site opted in. Chasing those AI citations seems like the only move that makes sense right now.

SEOgami

It is indeed an interesting experiment it is worth examining the halo effect in view of the fact that the visibility of AI may also have an effect on clicks for organic results in my opinion the effect will depend on the type of the niche and the intent of the search

check branded searches too. people might see you in AI first, then Google your site later.

Interesting test. The halo effect is probably the most important takeaway here. AI visibility may drive clicks beyond the citations themselves by increasing brand recognition and trust. I wouldn't opt out of AI features based on current evidence. Instead, I will track AI visibility, organic CTR, and conversions together.

That branded search point makes a lot of sense. If people spot the site in AI first and then search for it by name, the real lift might show up somewhere else entirely. Has anyone here actually split out branded vs non-branded clicks after a change like this?

that halo effect is real lol, opting out of ai overviews cost me clicks too, i wouldn't block the bots imo

I would compare the affected pages and search queries before drawing conclusions. A 21% drop could have several causes, so I would not assume the AI opt-out test is responsible without more evidence.

The halo effect is the part I'd watch. Being in the AI answer seems to lift clicks on the normal listing too, so AI clicks alone undersell it.