Zero-click searches: what they are and how to measure their effect on your site

A zero-click search is a search that ends on the results page: the searcher gets the answer from an AI Overview, a featured snippet, a knowledge panel or another results-page feature and never visits a website. On your own site, the effect shows up in Google Search Console as a specific pattern: impressions hold or rise, average position stays about the same, and CTR falls. You cannot win the click back on every query, so the response is to be the source the answer is built from, build demand for your brand, and focus effort on queries where the searcher still needs to visit a page.

Which results-page features cause zero-click searches#

Not every zero-click search is a loss. Someone who checks your opening hours in the local pack and then walks in is a customer who never visited your website. The problem is informational content whose business case depended on the visit.

How Search Console counts these searches#

Search Console counts an impression whenever a link to your site is shown on a results page the searcher sees, and a click only when they follow it. Google's documentation says a link inside an AI Overview counts as an impression, and that all links in an AI Overview share the single position the Overview occupies on the page. So a page cited in an AI Overview at the top can report a strong average position and very few clicks. As of October 2026, the Performance report has no filter that separates AI Overview impressions from other web results, so you infer the effect from the pattern, not from a dedicated metric.

How to see the effect in your own data#

1. 1

   Compare a period with the same period last year

   Performance → Date → Compare, last three months against the same three months a year earlier. Year over year removes seasonality, which otherwise looks exactly like a CTR change.

2. 2

   Remove brand queries

   Brand searches click at high rates and hide everything else. Filter them out with a regex, as in branded vs non-branded keywords.

3. 3

   Export queries with both periods

   Keep clicks, impressions, CTR and position for each period. Drop rows with fewer than about 100 impressions in either period; they are too noisy.

4. 4

   Flag the zero-click pattern

   Position changed by less than one place, impressions are flat or up, and CTR fell by a quarter or more. Those queries lost clicks without losing rankings.

5. 5

   Check the results page for the flagged queries

   Search a sample of them, or run them through our free AI visibility check, which shows the Google results, the AI Overview above them and the ChatGPT answer side by side with your domain marked. Note which feature now sits above you.

6. 6

   Group by query type

   Label each flagged query as definition, how-to, comparison, local or commercial. The losses usually cluster in one or two types, and that tells you where to change strategy.

Worked example#

Worked example (hypothetical numbers). A guide ranks for “what is a purchase order” at an average position of 2.3 in both periods. Impressions grew from 18,000 to 21,000 a month, but CTR fell from 11% to 4.5%, so clicks went from about 1,980 to about 945. A manual check shows an AI Overview with a definition, citing two other sites. The page did not lose a ranking: it lost the click to an answer on the page. The same site's comparison pages, “purchase order vs invoice” and “purchase order software”, held their CTR, because the searcher still needs to read and choose.

That split, definitions losing clicks while comparisons and commercial pages hold, is the most common shape of the effect. Measure it on your data rather than relying on a general industry figure: the share of zero-click searches varies by topic, country and device, and your own Search Console export is the only source that describes your queries.

What to do about zero-click searches#

• Be the cited source. Put a short, direct answer in the first paragraph, use headings that match the questions, and name your sources and dates. Anatomy of a page built for AI search shows the structure that answer engines can quote.
• Give a reason to click. A definition can be answered on the results page; a calculator, a template, original data, a worked example or a comparison table cannot. Add the part of the page that the summary cannot replace.
• Shift effort toward queries that need the visit. Comparisons, pricing, “best” lists, local services and product pages still send clicks. Weight your roadmap toward them.
• Build brand demand. A cited brand that searchers recognise gets searched by name later. Track branded clicks and impressions as a separate KPI.
• Track citations, not only clicks. For your most important informational queries, record whether an AI Overview appears and whether it cites you. AI Overview tracking covers how.
• Update your forecasts. If your CTR at a position has fallen while the position held, forecast with your recent curve, not an old benchmark. See CTR by position.

Which queries to keep chasing#

How to report it to a client or a manager#

Show non-brand clicks next to non-brand impressions and average position, so a falling click line with a stable position is visible at a glance. Add the list of queries flagged by the steps above, the feature now shown on each, and whether you are cited in it. The SEO KPIs page explains which metrics to keep in the report and which to drop. The search optimization layers framework is a useful way to explain why visibility in an answer counts even when it earns no click.

Go deeper#