If your Search Console report shows impressions rising while clicks stay flat, the useful question is what changed underneath the total. A page reaching more searches at low positions calls for a different response from a page losing clicks on the same searches near the top of the results.
AI Overviews and other answers on the results page are possible parts of the explanation. So are the searches being counted, their positions and the pages Google shows. I want to distinguish those possibilities before deciding that a page needs work—or that nothing is wrong.
What an impression measures
An impression records an appearance of a link to your site under Google's counting rules. For some result types, that can include a link the person did not scroll into view. It does not establish that they read the result, recognized your brand or found their answer. Google explains how impressions, clicks and position are counted.
A click records someone following a result to the site. Click-through rate divides clicks by impressions, so it can fall when impressions grow faster than clicks. Average position is an average across recorded appearances, not a fixed rank for the page.
The mix of searches matters. If a page begins appearing for additional searches far down the results, its total impressions can rise without a comparable rise in clicks. That does not establish whether the new searches are useful opportunities or whether the page answers them well. Read the searches and the page before making that call.
Shown far more often, clicked a little more
Where AI answers fit into the explanation
AI Overviews and answer-first interfaces can give people information without requiring a visit. Consider someone asking “what is search intent.” A summary might answer their immediate question; someone else might still want examples or help applying it.
Those are plausible ways a search could unfold. An impressions-and-clicks report does not tell you which one happened, whether an AI answer appeared, or why a particular person did not click.
Search Console's Generative AI performance report reports impressions across AI Overviews and AI Mode. Those impressions describe appearances in the supported features; they do not attribute a change in your site's clicks to an AI answer. A current result capture can help you decide what to change now, but it cannot reconstruct a historical search result.
I would use those views together: the performance report to locate a change, the relevant searches to understand the question, and an inspection of the page and results to test a possible explanation.
Why the page and query views matter
Start with the page whose traffic you are being asked to explain. Compare its searches over clearly stated date ranges, then distinguish searches it already appeared for from additional searches in the later period. Search Console omits some queries, so the visible query rows will not necessarily add up to the page total.
Look at position alongside impressions and clicks. A larger total made up of more low-position appearances is different from a click decline on an established search at a similar average position. Even the latter does not isolate a title problem or prove that a summary took the clicks; it identifies a more specific question worth investigating.
The Title Town PDX comparison below shows why I keep the measures together. Impressions and clicks both grew, but at different rates. It establishes that the click rate changed, without showing what caused the difference.
| Month | Google impressions | Google clicks | Click rate |
|---|---|---|---|
| April 2026 | 8,612 | 89 | 1.0% |
| May 2026 | 19,704 | 197 | 1.0% |
| June 2026 | 42,881 | 1,069 | 2.5% |
| July 2026 | 65,125 | 1,241 | 1.9% |
| August 2026 | 41,504 | 305 | 0.7% |
| September 2026 | 23,418 | 200 | 0.9% |
Could visibility still help the reader?
A useful answer can matter even when it does not produce an immediate visit. Someone might remember a source and return later, but aggregate Search Console counts do not establish brand recognition, authority or a later visit by the same person. Those require different evidence.
That distinction affects the work I would recommend. I would not remove a clear answer from a title merely because a report shows few clicks. I would first check whether the title accurately describes the page and helps the intended reader decide whether to open it. Keeping a useful answer visible is an editorial choice; the report alone does not prove its benefit.
What should the page offer beyond a summary?
For a person asking “how long does SEO take to work,” a short answer may be enough for the moment. Another person may need to understand what affects the timeline for their own site. The page should make that deeper help easy to find if it is the question the page is meant to answer.
That is why I look for a reader's next useful step: an example, a comparison, an explanation of tradeoffs or a way to apply the answer. I do not assume that a missing next step caused flat clicks, or that adding one will recover them. I check whether it makes the page more useful for the people it is meant to serve.
The distinction between informational and transactional pages helps here. A person learning a concept and a person choosing a product may need different answers and different ways forward.
How to choose what to do next
Use the report to narrow the decision rather than turn every change into the same fix:
| What you observe | What remains unknown | A useful next check |
|---|---|---|
| More impressions from additional searches at low positions | Whether those searches fit the page's purpose | Read the searches and compare their questions with the page's answer |
| Fewer clicks for an established search at a similar average position | What results appeared and what affected the choice to click | Inspect the current result and title before proposing a specific change |
| A page leaves a relevant question unanswered | Whether that omission explains historical clicks | Improve the identified answer, then measure what changes |
| A current title already gives a useful answer | Whether withholding it would help or harm the reader | Keep the reader's task in view instead of treating clicks as the only goal |
You can try this with two real reports in Impressions went up. Why didn't clicks?. The exercise asks whether to fund more pages and whether to change a stadium page's title, with the source rows and an explanation of each choice.
For your own report, write down the work you are being asked to approve. Then name the observation that supports it and the specific finding that could change your recommendation. That gives the next check a purpose.
Why are my impressions up but clicks down?
There can be several explanations, and the totals alone do not choose among them. Start by checking the reporting periods, the pages and searches behind the change, and their result positions. Treat AI answers, title problems and changing search needs as explanations to investigate where the evidence makes them relevant.
My priority is whether the page helps the people it is for. More impressions can point to a question worth examining, but they do not automatically justify more content, a title rewrite or a claim that the site is building authority. The recommendation should explain what you would change—or leave alone—and why.
First published on my old site February 7, 2026 and moved here November 1, 2026.