Impressions Up, Clicks Down? Three SEO Reports Worth Reading Carefully

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

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

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

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

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.

A test you can run: proposed protocol

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.

The practical takeaway

A diverging chart is a diagnostic clue. The useful response is a segmented investigation and a business-outcome check, rather than assuming either that SEO failed or that lost clicks have invisible value.

Sources and research notes

Sources reviewed on September 15, 2026. Public social pages and search extracts sometimes expose inconsistent relative dates; unverified publication dates are omitted. Reported results are attributed claims, not independently audited facts. Reposts of the same underlying campaign are not counted as additional experiments.

• Sean Johnson — LinkedIn — Indexed public post text.
• Jagadeesh J. — LinkedIn — Indexed public post text.
• Anant B. — LinkedIn — Indexed public post text.