Google AI Overviews and AI Mode get a 0.9 percent click-through rate, according to a small study cited in Search Engine Roundtable October 30, 2026, recap. The figure comes from an opt-out study, in which a site removal from AI features allowed comparison of behaviour with and without AI presence, and it lands alongside the Reddit referral warning as one of the starkest click numbers of the week.
A 0.9 percent click-through rate means that for every 1,000 AI answer impressions, about nine produce a click to a cited source. That is an order of magnitude below classic organic click-through rates for top positions, and it quantifies why publishers feel AI answers as a traffic loss even when their content is cited.
The study is small, and the recap presents it as such, so the figure should not be treated as a universal rate. Click-through varies by query type, by how many links an answer shows, and by whether the answer fully satisfies the query. But as a directional signal, 0.9 percent aligns with the Pew finding of 8 percent versus 15 percent click rates with and without AI summaries, and with the causal 38 percent suppression study: different methods, same direction.
AI answers are designed to satisfy the query inside the answer, so a click requires the user to want more than the summary provides. Most informational queries do not create that want, especially when the answer includes the key fact, figure, or definition. Commercial and exploratory queries, where the user wants to compare or buy, should click more, but the study average blends all query types together.
Link placement also matters. Inline links buried mid-paragraph, citations cards that some users never open, and answers that name a brand without linking it all suppress clicks even when the content is present. Google October 15 move to place inline links higher is, in this light, a response to exactly this problem, though without published click data its effect cannot yet be verified.
If clicks from AI answers are rare, the optimisation target shifts from click-through to citation and brand presence. Being named and linked in the answer builds familiarity that may convert later through branded search, even if the immediate click never happens. That is the brand-lift hypothesis from the opt-out study, and it deserves testing rather than dismissal.
The practical move is to measure the full funnel: AI impressions, branded search trends, direct visits, and assisted conversions, not just referral clicks from AI surfaces. A publisher that judges AI visibility by clicks alone will conclude it is worthless and opt out, potentially losing the brand presence that was quietly working.
A 0.9 percent click-through rate looks damning, but click-through was designed for a world where the result was a link, not an answer. Judging an answer surface by link clicks is like judging a television advertisement by store visits the same day: it captures one path and ignores the others, including memory, branded search, and later direct visits. The metric is not useless, but it is incomplete as a verdict.
Publishers need a fuller measurement habit: track AI impressions, branded search, direct traffic, and assisted conversions together, and look for patterns across weeks rather than single data points. A surface with low clicks but rising branded search is doing different work from a surface with low clicks and flat branded search, and only the fuller picture distinguishes them. The 0.9 percent figure should start that measurement, not end it.
Sources: Search Engine Roundtable daily recap, October 30, 2026.
Linkbaits helps you create content that earns citations and backlinks. Start with the free tools and build your first linkable asset this week.
Start free trial