Search Console Performance report: clicks, impressions, CTR and position

This guide is built from Search Console Help: the Performance report overview, About the data, Dimensions and data groupings, Advanced filtering and comparison, Troubleshooting data discrepancies and What are impressions, position, and clicks? Setting up Search Console is covered in our Search Console basics guide; this page explains what the numbers in the Performance report actually count.

The four metrics

The report opens on clicks and impressions for the past three months. Four metrics sit above the chart, and you can switch each one on or off:

• Clicks: "The number of times a user clicked your site from Google Search results."
• Impressions: "How many times your site appeared in Search results."
• CTR: "The click count divided by the impression count."
• Average position: "The average position of the topmost result from your site."

The details matter. An impression is generally counted when your link is on the current page of results, "whether or not the item is scrolled into view". Inside carousels and expanding results it usually has to be scrolled or expanded into view, and on infinite-scroll results such as mobile image search it has to scroll into view. Scrolling away and back does not count twice.

A click is counted when someone goes from Google to a page outside Google. Clicking the same result again after coming back "counts as only one click". Links that start a new Google search (Google calls them query refinements) get no clicks or impressions at all. If your site redirects the visitor after they land, the click still belongs to the URL Google showed.

AI features are counted with the same rules. A click on an external link in an AI Overview or in AI Mode is a click, standard impression rules apply, and "An AI Overview occupies a single position in search results, and all links in the AI Overview are assigned that same position." A follow-up question in AI Mode counts as a new query. Search Labs experiments are not included. More on how these features show your pages is in our AI Overviews guide.

How average position works

Position is counted from the top of the page, down the main column first and then the side column. Every result block takes one position, even a block with many links: Google's example has an AMP carousel in which every link sits at position 2, and a knowledge panel at position 6. Ads do not take a position, and neither do blocks whose only links start a new search.

For each query, only your topmost result counts. Google's worked example: one query shows your site at positions 2, 4 and 6, so it counts as 2; another shows it at 3, 5 and 9, so it counts as 3; the average is (2 + 3) / 2 = 2.5. A result that was never seen, for example on page 3 when the searcher stopped at page 1, records no position. That is why a comparison can show a dash instead of a number.

Google warns that position "can be misleading if you don't understand the subtleties". Position 11 could be the top of a knowledge panel on desktop, the first result on page 2, or a row of image results. Google's advice is to watch how position changes over time, especially sudden changes, and its use-case page goes further: "it's recommended to focus more on trends in impressions and clicks than on position alone." If you search for a query yourself and see a different position, that is expected, because results depend on time, place, device and search history.

Property totals versus page totals

This is the cause of most "the numbers don't add up" questions. Search Console counts data in two ways:

Google's example is a search that returns three pages from the same pet shop, and one user clicks all three. Counted by property, that is 1 impression, a CTR of 100% and a position of 1. Counted by page, each URL has 1 impression, a CTR of 33% and a position of 2, because (1 + 2 + 3) / 3 = 2. So when several of your pages appear for the same searches, CTR and position look better at property level than on the Pages tab.

Data is also assigned to the canonical URL. When someone clicks a duplicate URL in the results, "the click counts toward the canonical URL, not the URL the user visits". Mobile and desktop versions of a page are merged in the same way. Our canonical tags guide explains how Google chooses that URL, and the URL Inspection tool shows which one it picked.

Dimensions: queries, pages, countries and more

The tabs above the table group the data by queries, pages, countries, devices (desktop, tablet or mobile), search appearance or dates. A few rules from Google's dimensions page are useful to know:

• Queries: query filters are not case-sensitive. For multimodal searches (Google Lens, Circle to Search and other image-based searches) there is no query text, so the Queries tab is not available for that search type.
• Branded and non-branded: a query filter splits searches that include your brand name from those that don't. Data starts from March 2025, it isn't offered for sites with few impressions or for sub-properties such as a /blog/ folder, and Google says some queries "might be incorrectly identified". What the related Insights report does with this split is in our Search Console Insights guide.
• Pages: the final URL after any redirects, normally the canonical.
• Countries: where the search came from.
• Search appearance: result types such as review snippets, product snippets, merchant listings, videos and translated results. Only types your site has impressions for appear in the list. Filtering by a type shows that the searcher saw your URL with that feature, but the click may have been on a plain link to the same URL, because clicks are assigned to a URL, not to a URL and a feature.

Filters, search types and comparisons

The date and search type filters are always there and can't be removed. Search types are Web (split into text-based and multimodal), Image, Video and News. Each one is reported separately, and Discover and Google News have reports of their own (see our Google Discover guide).

You can stack filters, for example country = USA and device = mobile, or click any table row to filter by it. Query and page filters work in three ways: a list of exact items, "containing" or "not containing" a string (not case-sensitive, otherwise exact, spaces included), or a regular expression. The regex option is covered in detail in our Search Console regex guide.

Compare puts two values side by side: two date ranges, two countries, two devices, two search types or two search appearances. The table adds a Difference column you can sort by. Only one comparison can run at a time. Google suggests weekly or monthly granularity when comparing dates, so that a Monday isn't compared with a Saturday. When one side of a comparison is rare, its row shows ~ (not available); filtering by that value shows the numbers.

You can mark events on the chart with annotations, such as a redesign going live or a fix to the site. Google's annotations page allows up to 200 per property, 120 characters each. Notes older than 500 days are deleted automatically, and annotations don't show in comparison mode.

Why rows are missing

Three limits mean the table never holds everything:

• Anonymized queries. Some queries are left out to protect privacy. A Search Central blog post on data filtering defines them as queries "that aren't issued by more than a few dozen users over a two-to-three month period". They are never listed in the table but are counted in the chart totals, unless you filter by query.
• The 1,000-row limit. The table shows at most 1,000 rows, and the interface exports at most 1,000 rows. The same post puts the Search Analytics API (and the Looker Studio connector) at up to 50,000 rows per day per site per search type.
• Data truncation. Google says Search Console "stores and shows only the most important data rows", and the most complete list of queries comes from the bulk data export to BigQuery. That export needs a Google Cloud project with billing and only starts from the day it is set up; for earlier dates Google points to the API or the reports.

The practical result: adding up the query rows won't match the chart total, and a "containing" filter plus a "not containing" filter for the same word won't add up to the unfiltered total either. The table may also disappear when it would add nothing to the chart. In an export, values shown as ~ or - become zeros.

Dates, time zones and fresh data

Daily data is in Pacific Time in every view except the 24-hour view, which uses your browser's local time and shows hourly points. Weeks run Sunday to Saturday and months are calendar months, so a range that starts mid-week gives a partial first point.

The newest data can be preliminary and change over the next few hours; the chart draws it with a dotted line. By default the report shows complete days only, and preliminary days appear when you pick them in the date filter. Google's troubleshooting page says collected data "is usually available in 2–3 days".

Why it differs from Google Analytics

Search Console and Google Analytics count different things, so their numbers will not match. Google lists the reasons: privacy filtering, processing (for example removing duplicates), the 2–3 day lag, Pacific Time days, and that tools such as Google Analytics only track visitors who have JavaScript enabled. A Search Console click is a click on a result; a GA4 session is a visit that loaded your tag, which also depends on consent. Our guide to linking Search Console to GA4 shows how to put both sets of numbers into one report.

Putting the report to work

Google's common tasks page suggests a few uses: sort the Queries tab by clicks to see what brings people in; sort the Pages tab by CTR, lowest first, and consider updating the title and description of pages that are seen but rarely clicked; click a query to see which of your pages Google showed for it; and compare periods to see whether a change on a page lines up with a change in search traffic.

What follows is our own routine, not Google's. Once a month we compare the last 28 days with the 28 before, by week, and sort by the Difference column on both Queries and Pages. We note site changes as annotations on the day they go live. We look at pages with many impressions and a low CTR first, since a better title is the cheapest fix on the list. We read position as a trend per page, not as a ranking. And we report Search Console clicks next to GA4 enquiries, never added together; our monthly website report guide shows the layout. For a quicker summary of what is rising and falling, Search Console Insights puts it on one screen.

If you would like a hand reading your own data, our website team can review your Search Console property with you.

Frequently asked questions

What is average position in Google Search Console?

It is the topmost position of your site (or of a page, on the Pages tab) in Google results, averaged over all impressions. Positions count from the top of the page, every result block takes one position, and results that were never seen record no position.

Why don't the table rows add up to the chart total?

The chart includes anonymized queries and rows beyond the 1,000-row table limit, and it counts by property while the Pages and Search appearance tabs count by page. Google lists these as the usual causes.

How is CTR calculated in Search Console?

Clicks divided by impressions. It is usually higher at property level than per page when several of your pages appear in the same results, because the property counts one impression for all of them.

How fresh is Search Console Performance data?

Collected data is usually available in 2–3 days. The newest days can be preliminary, shown with a dotted line, and may change over the next few hours. Days are in Pacific Time except in the 24-hour view.

Are AI Overviews and AI Mode included in the Performance report?

Yes, in the Web search type. A click on an external link in an AI Overview or AI Mode counts as a click, standard impression rules apply, and all links in an AI Overview share its single position.

For more on Search Console, see our guides to the Page indexing report, Links report, Crawl Stats and Looker Studio dashboards.