AI Overviews, Impressions and Clicks: Interpret the Gap

AI Overviews, impressions and clicks are often discussed as if a rising impression count and falling CTR prove one cause. They do not. Three SEO reports show different patterns, including a case where clicks grew despite a lower CTR. This review separates the arithmetic from evidence about the search experience.

The lesson

Investigate clicks, impressions, query mix and search-result changes together. A lower click-through rate does not by itself identify the cause or the business impact.

What people reported

These public accounts describe different setups. Read each reported outcome with its design limits; repeated descriptions of the same campaign are not independent replications.

Sean Johnson: More impressions alongside fewer clicks

Sean Johnson reports a six-month comparison across properties he manages: impressions increased 26% while clicks fell 32%, even as average position improved. He explicitly presents this as anecdotal evidence rather than proof of AI Overview impact. The underlying query and page distributions are not available for independent analysis. Read the original report: Sean Johnson — LinkedIn

The reported pattern is compatible with several mechanisms, including more exposure on low-click queries or changing search features. Without query-level exposure data and a suitable comparison, it cannot identify the share of decline caused by AI Overviews. Improved average position can also reflect a changed mix of queries.

Jagadeesh J.: Improved visibility without corresponding click growth

Jagadeesh J. describes a client whose impressions and rankings improved year over year while Google/Bing clicks declined or stayed flat in some areas. His observation helped motivate exploring AI referrals. It does not show that the lost clicks came from AI Overviews or that another channel replaced them. Read the original report: Jagadeesh J. — LinkedIn

Visibility and visits measure different stages. Separate existing queries from newly appearing ones before deciding whether the established audience lost interest. More impressions on broad informational searches can dilute aggregate CTR even if important commercial pages remain stable.

Anant B.: A lower CTR while total clicks increased

Anant B. reports impressions rising from roughly 298,000 to 2.13 million and clicks from around 3,000 to 8,640, while reported CTR fell from about 1.0% to 0.4%. This is a useful counterweight: impressions outpaced clicks, but clicks did not decline. The post’s broader AI explanation is not established by those totals alone. Read the original report: Anant B. — LinkedIn

This example directly challenges the assumption that lower CTR means fewer visits. Its reported clicks increased while impressions grew faster. Evaluate the count of useful visits and conversions first, then use CTR to investigate the exposure mix rather than treating the percentage alone as a success or failure verdict.

What the experiences have in common

The similar experience is the weakening relationship between impression growth and click growth. The third report prevents an easy misreading of the first two: a falling ratio can coexist with substantial traffic growth. CTR is clicks divided by impressions, so both numbers matter.

Changes in the query mix also matter. A site gaining many low-position impressions for new queries can show a lower overall CTR without losing its existing audience. New result features, ads, seasonality, device changes, measurement changes or altered demand may contribute. An average position aggregates different queries; it is not a guarantee that the same results held the same places.

What these reports cannot establish

These are account-level comparisons, not experiments that turned AI Overviews on and off. Their time windows and business contexts differ. Some authors offer causal explanations more confidently than their evidence allows. The articles’ observed trends should be retained while those explanations remain hypotheses.

Separate CTR arithmetic from causal attribution

CTR is clicks divided by impressions, multiplied by 100. In a hypothetical example, 1,000 clicks from 50,000 impressions is 2%; 1,200 clicks from 100,000 impressions is 1.2%. Clicks rose by 20% despite lower CTR. Google's Search Console metric definitions explain how counted exposure differs by result type.

An August 2026 preregistered search field experiment, reported as an arXiv preprint, studied 1,100 participants and found higher publisher click-through when AI features were removed. That experimental evidence is stronger for causal interpretation than an account screenshot. It still does not quantify why a particular site's clicks changed.

A test you can run: proposed protocol

Use the following protocol as a starting design. Choose one outcome and a practical review window before making changes, and retain the original observations so a disappointing result remains reportable.

1. Export comparable periods with page, query, country and device dimensions. Document known reporting breaks, major releases and seasonal events.
2. Separate previously observed queries from newly gained queries. Compare branded, informational and commercial groups rather than only the site total.
3. For important queries, record live result features and repeat observations. Keep a comparable query group without the observed feature where possible.
4. Compare changes in clicks and CTR within similar query and position groups. Track qualified leads or purchases alongside traffic.
5. Report what the data can establish: the gap, the affected segments and the timing. Describe an AI Overview explanation as an association unless the design supports a causal claim.

Conclusion

Rising impressions and falling CTR are a diagnostic starting point, not proof that AI Overviews caused a traffic loss. The accounts include both falling and growing click totals. Current experimental research supports a possible effect of AI search features, but site-specific attribution still requires better measurement.

Compare like-for-like page, query, country and device segments across comparable windows. Separate new exposure from established queries, record major changes and inspect qualified outcomes. Prioritize pages losing valuable visits; a diluted percentage alone is not a reason to rewrite content that is still serving readers well.