Google Search Console Performance Report Explained

The Search Console performance report shows how your site appears in Google Search through four metrics: clicks, impressions, click-through rate and average position. You can split them by query, page, country, device, search appearance and date. Read it by comparing two periods and filtering to one page or query group, never by staring at site-wide totals.

Context matters when you read click data: SparkToro’s 2024 study (with Datos clickstream data) found that 58.5% of US Google searches and 59.7% of EU Google searches ended with zero clicks, so plenty of impressions without clicks is normal.

What the performance report actually measures

The report is built from Google’s own logs of what appeared in search results and what people clicked. It is not an analytics tool in the GA4 sense. It cannot tell you what a visitor did after landing on your site, and it does not count visits from other search engines. What it can tell you, better than anything else, is how you are showing up before the click.

Google defines the four metrics in its Performance report documentation. Here is how I explain them to clients and in my own audits.

One detail that trips people up: a result can count as an impression even when it was lower on the page and the person never scrolled to it, depending on the result type. Google explains the counting rules in its guide to how Search Console counts impressions, clicks and position. That means impressions are a measure of opportunity, not attention.

The six dimensions and what each one answers

Dimensions are the tabs under the chart. The metrics stay the same; the dimension changes how the rows are grouped. Each one answers a different question, and mixing them up is the cause of most bad conclusions.

Two things to remember. First, the tables aggregate differently: queries, countries, devices and dates are grouped at property level, while pages and search appearance are grouped at page level. Second, click and impression data for the variations of a page is assigned to the canonical URL that Google selects, so a page’s numbers may sit under a different URL from the one you expected. If a URL is missing from the Pages tab, check which canonical Google chose with the URL Inspection tool.

How to read average position without fooling yourself

Average position is the most misread number in SEO reporting. Google states that the position shown is the topmost position occupied by a link to your property or page, averaged across all queries. Three consequences follow.

• It is an average across many searches. A page that ranks first for one query and twentieth for another shows a middling number that describes neither.
• It changes when your mix of queries changes. If a page suddenly appears for hundreds of new long-tail queries at low positions, its average position gets worse while its real visibility has grown. This is good news that looks like bad news.
• It depends on the filter. Position is only meaningful once you have filtered to a single query, or to a single page and a tight group of queries.

My rule: never report site-wide average position as a headline. Use it at query level, or at page level after filtering to the one query you care about. For a ranked list of what to do with the pages sitting just off page one, see my guide to striking-distance keywords in Search Console.

Filters and comparisons that make the data usable

The raw report is too noisy to act on. Filters and comparisons are what turn it into answers. Google documents the options on its advanced filtering and comparison help page, and the useful points are these.

• Query and page filters. You can include or exclude by “contains”, and you can use a custom regular expression for more precise matching. My Search Console regex filter examples cover brand versus non-brand, question queries and URL folders.
• Search type. Web results are the default view. Image, video and news results have their own data, so switch the search type before you compare numbers with last month.
• Date range and granularity. You can aggregate by day, week or month. Weekly or monthly granularity smooths out the day-of-week pattern that makes daily charts jagged.
• One comparison at a time. You can compare two periods, two countries, two devices, two pages or two search appearance types, but only one comparison is active at once. Adding a new one replaces the old one.

Comparing periods is the single most useful habit. Compare the last 28 days against the previous 28 days, or against the same period last year if your business is seasonal. A drop that looks alarming on its own is often just the same dip you see every year.

Data limits you need to know before you trust a number

The report is accurate enough for decisions, but it is not a complete record of every search. Google’s explanation of Search Console data coverage sets out the limits that matter.

• Anonymised queries. Google may not track queries that are made a very small number of times or that contain personal or sensitive information. Those queries still count towards your chart totals, but they never appear as rows.
• A row limit. Tables show a maximum of 1,000 rows, so some rows are omitted. For a large site, filter to a folder or a page group to see more of the picture, or use an export route that suits your data volume.
• Totals can exceed table sums. Because of the two points above, the chart total can be higher than the sum of the rows beneath it. This is expected, not a bug.
• Preliminary data. The most recent days are marked with dotted lines and can change within hours. Do not draw conclusions from the last day or two.

The practical lesson is to judge trends using the chart and the totals, and to use the table rows for finding examples. Do not rebuild the chart by adding up rows.

Five analyses to run every month

Here is the routine I would run on any site. It takes about an hour once you are used to it, and each step ends with a decision, not just an observation.

1. Brand versus non-brand clicks. Filter queries to exclude your brand name and compare the last 28 days with the previous 28. Brand clicks measure demand you already have; non-brand clicks measure how well SEO is bringing in new people. If you only look at the total, brand growth can hide a non-brand decline.
2. Pages with high impressions and low CTR. Sort the Pages tab by impressions, then look for pages with a CTR well below others of similar position. Those are title and snippet candidates, or pages where a result feature is pushing you down.
3. Pages that lost the most clicks. Use the date comparison on the Pages tab and sort by the change in clicks. For each of the top decliners, note whether the loss is from fewer impressions (visibility), a lower position or a lower CTR. Each points to a different fix, and the traffic drop diagnostic checklist shows how to follow it up.
4. Device and country splits. Compare mobile with desktop and your top three countries with the previous period. A decline on one device only usually points to a technical or layout issue; a decline in one country can be seasonal or a result of a competitor.
5. New queries. Filter to the last month and look for queries that were not present before. These tell you what Google already associates your pages with, and they feed your content plan.

Write down the numbers and your decision each month, even in a plain spreadsheet. Once you have a few months of notes, a dashboard becomes worthwhile; I walk through that in building an SEO dashboard in Looker Studio, and choosing which numbers deserve a place on it is covered in SEO KPIs that actually matter.

Common misreadings to avoid

• Treating impressions as visits. An impression means a link was shown, or potentially seen. It does not mean anyone read your title.
• Chasing CTR in isolation. A low CTR on a page ranking fifteenth is normal. Judge CTR against position and against the query, since a navigational query behaves differently from a research query.
• Comparing mismatched periods. A 28-day window against a calendar month, or a holiday week against a normal week, produces fake changes.
• Ignoring the property type. If you verified a domain property and a URL-prefix property, the numbers can legitimately differ, because they cover different sets of URLs.
• Using the report as a rank tracker. It reports what actually happened to real searchers, which is better than a rank tracker for decisions but worse for checking one keyword on one day.

If you are also tracking results beyond Google’s own data, how to track SEO performance shows how to combine this report with analytics and business outcomes. Together they tell the full story from impression to enquiry.

Where the performance report fits in an audit

When I audit a site, the performance report is the first thing I open, because it shows what Google already thinks the site is about. I compare it with the pages the business wants to rank, and the gaps become the first findings. It is one input among many, and the full list of what a professional SEO audit includes shows where it sits next to crawling, indexing, content and links.

Questions people ask about the performance report

If you would like a second pair of eyes on your own numbers, I offer a free audit as part of my SEO audit services. Get in touch and send me the site you want reviewed.