Search Console's Performance report

Almost every conversation about organic search performance ends up in the same place: a chart from Search Console. It is the main first-party source Google provides, and it is free, so it deserves to be used. It also deserves to be read carefully. The numbers are counts of events in Google Search, defined by Google, aggregated in particular ways and subject to known gaps. This article sets out what the documentation says each number means and where the limits are.

What impressions, clicks and click-through rate count

Google’s page on impressions, position and clicks defines the three basic measures. An impression means a user has seen, or potentially seen, a link to your site. In general it is counted when an item appears in the current page of results, whether or not it was scrolled into view, provided the user did not need to click to reveal more results. Inside carousels and expandable sections, the item usually has to be scrolled into view or expanded. Scrolling or paging away and back does not count a second impression.

A click is counted when a user follows a link that takes them to a page outside Google Search, Discover or News. Clicking through, returning and clicking the same link again counts once. Clicks on links that stay inside Google, such as a query refinement that runs a new search, are not counted as clicks or impressions for that link. Click-through rate is simply clicks divided by impressions.

Two consequences follow. Google itself says the aim should be meaningful impressions, not simply more of them. And the report records what happened in Google’s results, not what happened on the site afterwards. Sessions, conversions and revenue come from other systems, and the Performance report cannot stand in for them.

Why average position needs care

Position is calculated only for Google Search results. For each query, the report takes the topmost position your property occupied, then averages those values across the queries where it appeared. Google’s own worked example: topmost positions of 2 and 3 across two queries give an average of 2.5, even though the property also appeared lower down in both.

Google is direct about the weakness. It describes position as a complex metric that can be misleading, notes that a position number can mean different things in different situations (a top-right knowledge panel entry and the first item on page two can share a value), and says to monitor change over time, especially sudden change, as well as the absolute figure. It also explains why your own search may not match the report: the value is an average across all searches, and results vary with search history, location and other variables.

Why the chart and the table do not always agree

The Performance report overview explains the aggregation rules. In the chart, data is aggregated by property: if two results from your site appear for one query, that counts as a single impression. In the table, grouping by queries, countries, devices or dates is also aggregated by property, while grouping by pages or search appearance is aggregated by page. Google says differences between chart and table totals are usually down to this.

Data is also assigned to the canonical URL Google selects for a page, not necessarily the URL a visitor saw. Desktop and mobile variants, parameter variants and in some cases alternate language versions can be combined under one URL, and a redirect after landing does not change which URL receives the click. If a page’s numbers look wrong, the first question is which URL Google treats as canonical. Our article on audit findings that get implemented covers how such findings are written up, and the canonical question is one reason a report and a crawl can disagree.

Settings matter too. Weekly and monthly granularity smooth out weekends and holidays, while daily helps spot issues or spikes. The newest data can be preliminary and may change within hours, which Google marks with a dotted line. Exports turn values shown as unavailable into zeros, a trap we cover in unavailable is not zero.

Logging errors, product changes and the anomalies page

Google publishes a data anomalies page for events that can affect report data, such as a change to aggregation methods or a logging error. It records known issues from roughly the last 3 to 16 months. The entries are specific. Recent entries include an error that affected how impressions were reported over a long period while leaving clicks unaffected, and two days missing from the bulk data exports for some properties that Google says will not be recovered. A further entry notes that FAQ rich results stopped appearing in Google Search from 7 May 2026, which reduced reported FAQ impressions.

Google’s guidance on debugging traffic drops tells readers to check this page, because a drop may come from a change in data processing or a logging error. We would go further as a working habit: when a number moves sharply for no obvious reason, look here before commissioning any work.

Keeping history beyond the 16-month view

The traffic-drop guidance suggests selecting the last 16 months of data so a drop can be seen in context and compared with the same period a year earlier. For anything longer, it points to the Search Analytics API or bulk data export to pull the data into your own systems.

The bulk export page describes scheduling a daily export of performance data to BigQuery, where it can be queried or sent on to external storage. It says you will see all the performance data available to Search Console for the property, with the exception of anonymized queries. That exception matters: an export is not a complete list of every query behind the totals, and the pages we fetched do not give more detail than that. The anomalies page also shows an export can have gaps of its own. If a site’s history will be needed in two or three years, the practical point is that it has to be stored starting now, because the report will not hold it for you.

What the report cannot tell you

Some limits are not about counting at all. The report cannot say why a result changed. Google also states that results are specific to the time, place, device and recent history of the person searching, so a query in your list may not show your site when you run it yourself. Explaining a change takes comparison and context, which is why the traffic-drop guidance combines the report with Google Trends and other Search Console reports.

Where this fits at Cultured Digital: measuring search

Measurement is a group of work within our SEO Strategy service, set out in the Measure guide: measurement design, reporting that answers business questions, and ongoing strategic consultancy. It is also the fifth step in how we work: check whether the work produced the intended result. The limits above are why we want reports to state what is known and what is not. We do not promise rankings or traffic numbers; we set out the opportunity and the evidence, and measure against it. Our articles on what clients should see in SEO reporting and on unavailable data develop the reporting side.

What Search Console’s Performance report can and cannot tell you