Lots of impressions, almost no clicks in Search Console: what to check, in order

What does Search Console count as an impression?

An impression is counted when a link to your site is on the page of results someone loaded, whether or not they scrolled down to it. So an impression means "was on the page", not "was read and turned down".

Inside a carousel the item usually has to be scrolled into view, and a result on page 3 counts nothing unless someone opens page 3. Three counting rules explain many puzzling rows. Every link inside an AI Overview gets that block's single position. An image of yours shown in the main results counts an impression for the page it sits on, and expanding the thumbnail is not a click. And the chart counts one impression per search for your whole site, while the Pages tab counts each of your URLs separately, so the two totals rarely match.

Work in the Performance report for Search results, search type Web, over the last three months, with the four metrics switched on: Clicks, Impressions, CTR and Average position.

In what order should you check the causes?

Check the causes that call for no action before the one that does: is there enough data, where do the impressions sit, who is searching, for what, and what did the results page show. Only then compare a query's CTR with your own site at the same position.

The table gives the order, the view that shows each cause and the usual verdict.

Are your impressions coming from page 2 and beyond?

Many impressions without clicks come from there. Average position is the average of your topmost position across all impressions, so a page at 9 can be a mix of position 3 and position 30. Rows beyond about position 10 tend to collect impressions and few clicks, and rewriting their title does not move them up.

1. Open the Queries tab, switch on Average position and sort by Impressions.
2. Export the table and, in a spreadsheet, put each row in a band: positions 1 to 3, 4 to 10, 11 to 20, beyond 20. Add up impressions and clicks per band.
3. If most impressions sit beyond 10, the low CTR is a position effect. The work, if any, is ranking: a fuller answer, internal links, the right page for the query.

Queries averaging around 8 to 15 deserve a look: they can be on page one part of the time. A query averaging 40 rarely justifies a rewrite on its own.

Are brand searches distorting the click-through rate?

People who type your name click on you; people who type a generic question compare. Mixed together, a few brand queries can hide a weak non-brand CTR, and on a site nobody searches by name the opposite holds. Split the two before reading anything else.

Search Console has a branded and non-branded queries filter that separates searches containing your brand name. Google notes that it is not available for sites with a low number of impressions and that some queries may be classified wrongly.

Without it, use + Add filter, choose Queries, then Queries not containing and your brand. For several spellings, choose Custom (regex), enter them as (yourbrand|your brand|yourbrnd) and pick Doesn't match regex.

On a young site, most clicks are often brand clicks. That shows people who know you can find you; it says nothing yet about how the site does on generic searches.

Is the page shown for searches it does not answer, or competing with your other pages?

Click a page's row in the Pages tab, then open the Queries tab: you see every search that showed that page. Someone looking for a login screen or another company, a search in another language or with a different intent produces impressions that no title will turn into clicks.

For an off-target query, either let it go or, if it matters to your business, give it a page built for it. Check first that the query can be won.

The reverse check finds cannibalisation: click a query's row, then open the Pages tab. Two of your URLs on one query is not always a problem, since the chart counts your site once per search. It becomes one when impressions alternate between the URLs from week to week and neither holds a steady position. Then merge them, or make one clearly the answer and link to it from the other.

• Odds Maker: reads the real first page for a query and says whether a new page is worth writing

Are AI Overviews and other zero-click results taking the clicks?

An AI Overview can answer a search before anyone clicks. Search Console cannot show that query by query, but it now shows AI impressions page by page. Impressions in AI Overviews and AI Mode are counted inside the Web search type of the Performance report, and no filter there isolates them.

The Generative AI performance report, rolled out to all websites as of 31 August 2026, shows those impressions by page, country, device and date, without queries or clicks.

1. Open the Generative AI performance report, choose the Pages tab and note the page's impressions for a date range.
2. Open the Performance report for the same dates, click the page's row and note its total impressions.
3. If a large share comes from AI features and the page's queries are questions with a short factual answer, the low CTR comes from the format of the results page, not from your title.

Images behave the same way: a thumbnail of yours in the main results counts an impression for its page, and opening it is not a click. The Search appearance tab lists the result types Google reports for your site, such as Videos or Merchant listings.

When not to act: a definition or a quick fact answered on the results page will not become a click because the title changes. Keep the page if it serves other purposes, and spend the effort where searchers still need to visit.

Is the CTR low for that position on your own site?

This is the only cause a better title and description fix. Compare each query with your own site's other queries at the same position, not with a published industry curve: your result types, brand and audience set your curve. A query far below its neighbours, on a topic the page does answer, has a snippet problem.

Google's own help suggests the Queries dimension for this: queries with high impressions and low CTR, to improve the page title and description.

1. Apply the non-brand filter, keep the last three months and export the Queries tab.
2. In a spreadsheet, group the rows by rounded average position: 1, 2, 3, 4 to 5, 6 to 10.
3. For each group, divide total clicks by total impressions. Do not average the rows' CTRs, or a query with 20 impressions weighs as much as one with 20,000.
4. Mark the queries with real impressions whose CTR is well below their group's, for example under half.
5. Search each one yourself. Does your title use the searcher's words, say what the page gives, and fit the type of result Google shows there?
6. Rewrite the title and description of the page that ranks, note the date, and compare two equal periods afterwards.

• Click Radar: reads your Queries and Pages exports, splits brand from non-brand, finds CTR gaps against your site's own curve and says when the sample is too small to conclude

When is it better not to act?

When the sample is too small, the data is incomplete or the cause is one no snippet fixes. A young site gets few non-brand impressions per query, a change read on a few days or on preliminary data is noise, and positions beyond page one, AI answers and off-target queries are not title problems.

For scale: tackplus.app, a second site by the author of this guide, had 57 clicks from 3.86k impressions between 15 September and 12 October 2026, a CTR of about 1.5%. Once numbers that size are spread across queries, a single query moving a few positions can change the picture. At that size, watch the weekly trend and rewrite nothing yet.

A bigger sample is no benchmark either. macplus.pro, a new domain run by the same author, showed 4.47k clicks from 59k impressions (CTR 7.6%, average position 7.9) between 14 July and 12 October 2026. That total mixes brand and non-brand searches at every position: it is where the split starts, not a target for another site.

Compare complete days only (preliminary data shows as a dotted line), equal periods, and Weekly granularity, which Google suggests when comparing date ranges. Change one thing at a time and write down the date.

• Guide: Discovered - currently not indexed: when pages never reach the results in the first place
• Index Doctor: reads the Page indexing export and ranks the cheapest fixes by page template

If impressions are close to zero, not just clicks, the page may not be indexed at all. That is a different diagnosis.