Read the Search Console Performance Report: Clicks, CTR and Position

The search console performance report is the page most site owners open first and understand least. It shows four numbers (clicks, impressions, click-through rate and average position), and each one has a precise definition that is easy to misread. A rising average position can mean you improved, or it can mean you started appearing for many new, low-ranked searches. A table that does not add up to the chart above it can look like a bug, but it is how the report is built.

This guide is for site owners, marketers and developers who have Search Console set up and want to read the data honestly. You will learn what each metric counts, why totals sometimes disagree, how filters change the picture, and how to export your data so you keep a history beyond what the interface holds. If you have not added your site yet, start with Set Up Google Search Console.

Before you begin

You need access to a verified Search Console property for your site. Pick a property that matches how people reach you; if you are unsure, read Domain Property vs URL Prefix first, because a URL-prefix property only shows data for URLs that begin with its exact prefix.

Allow about 30 minutes for the first pass. Nothing here changes your website, so the risk is low. The one thing worth doing early is deciding where you will keep long-term history. Google's Analytics help states that Search Console keeps data for the last 16 months, so older rows disappear from the interface. Step 6 shows two ways to keep a copy.

How the Search Console performance report counts its metrics

Google defines these in its own help pages, and the exact wording matters.

Impressions. An impression is counted when a user has seen, or potentially seen, a link to your site in Search, Discover or News. For items in carousels or expandable sections, the link generally has to be scrolled into view or expanded to count. Scrolling away and back during the same query does not add impressions.
Clicks. A click is counted when a user goes from Google to a page outside Google Search, Discover or News. Clicks that stay inside Google, such as refining the query, are not counted. If someone returns and clicks the same result again, it is still one click.
CTR. Click-through rate is clicks divided by impressions.
Average position. Position is calculated only for Google Search results. For each query, the reported value is the topmost position your link held. Those values are then averaged. If your best position was 2 in one query and 3 in another, the average is 2.5. Positions are recorded only for results that received an impression, and a dash means no position was recorded.

One more rule affects every table: data is assigned to the canonical URL that Google selects for a page, so desktop and mobile versions of the same page are combined. You do not need to merge them yourself. If you want to understand canonical choices, see Fix Duplicate URLs and Incorrect Canonical Tags.

Step-by-step instructions

Step 1: Set the search type and date range first

Open Search Console, choose your property, and open the Performance report for Search results. The default view covers the past three months. Before reading any number, check two filters above the chart.

The search type options are Web (text searches), Web with multimodal searches such as Google Lens or Circle to Search, Image, Video and News. Each type has its own clicks and impressions, so a drop in one type is not a drop in another. The date filter also controls granularity: hourly in the 24-hour view, and daily, weekly or monthly otherwise. Weekly and monthly views smooth out weekday and holiday swings, which helps when you are judging a trend rather than a single day.

The newest data may be preliminary and can change within a few hours. It appears as a dotted line on the graph. Do not draw conclusions from the dotted part.

What success looks like: You can say exactly which search type and period each number describes.

Step 2: Turn on all four metrics and read the tabs

Click the metric tiles so clicks, impressions, CTR and average position are all active. Then work through the tabs under the chart: Queries, Pages, Countries, Devices, Search appearance and Dates.

Queries, Countries, Devices and Dates are aggregated by property. Pages and Search appearance are aggregated by page. Google explains that this difference can make chart totals differ from table totals. For example, if one search result element contains several links from your site, a property-level count treats it as one impression, while a page-level count gives each URL its own.

What success looks like: You know which tab is property-level and which is page-level before comparing totals.

Step 3: Understand what the report leaves out

Two limits shape the data, both explained in a Google Search Central article on performance data filtering.

First, privacy filtering. Queries that are not issued by more than a few dozen users over a two-to-three month period are called anonymized queries. They are left out of the query table, but they are included in the chart totals unless you filter by query. So if you add up the clicks in the Queries table and get less than the chart shows, nothing is broken. The difference is clicks from anonymized queries.

Second, applying a query filter removes anonymized queries entirely. If you filter to queries containing "shoes" and then queries not containing "shoes", the two results will not add up to the unfiltered total.

There is also a daily row limit. The interface exports at most 1,000 rows. The Search Analytics API and the Looker Studio connector can return up to 50,000 rows per day per site per search type, although that ceiling is not always reached. Requests that do not involve query or URL dimensions, such as countries, devices and search appearance, show and export all the data.

What success looks like: You expect small gaps between row sums and chart totals and know why they appear.

Step 4: Compare pages and queries fairly

Raw numbers mislead when you compare unlike things. A page at position 1 will usually have a higher CTR than a page at position 8, so a "low" CTR on a deep position is normal. Compare pages at similar positions, similar search types and similar devices. Use the Devices tab or a device filter to split mobile from desktop, and use date comparison to line up two equal periods rather than one week against a month.

A practical review has three passes:

Sort Pages by impressions and look for pages with many impressions and few clicks. The search result may not be convincing. A clearer title and description can help, as covered in Write SEO Title Tags and Meta Descriptions That Match Search Intent. It may help, but nothing here is certain.
Sort Queries by clicks and note which searches bring the most visits. These are the topics to protect when you edit or redesign.
Look at Countries and Devices for surprises, such as a sudden change in one device type.

What success looks like: You have a short list of pages and queries to investigate, each tied to a specific number.

Step 5: Read average position with care

Average position is the most abused number in the report. It is an average across queries of the topmost position, so it moves when your mix of queries changes. If a page starts earning impressions for dozens of long, specific searches where it sits on page two, its average position gets worse even though nothing got worse for your best queries.

To see a meaningful position, filter to one page and one query, or to a small group of related queries, and look at the Dates tab. Position is a clue about visibility. It is not a measure of quality and not a prediction of clicks.

What success looks like: You quote position only for a defined page and query set, never as one site-wide score.

Step 6: Export and keep your own history

Use the Export button to download the chart and table data. Cells shown as "~" or "-" in the report export as zeros, so do not read zero in a spreadsheet as a measured zero.

For a durable archive, Search Console offers a bulk data export to BigQuery under Settings, then Bulk data export. It needs a Google Cloud project with BigQuery enabled and permission granted to a Search Console service account. The first export happens up to 48 hours after a successful setup, and it does not backfill: data from before setup has to come from the API or the reports. Data accumulates in your project until you set a partition expiration, and if you do, keep it at 14 days or longer. Storage and query usage above the free level is charged by Google Cloud, so set a budget alert.

What success looks like: You have either a dated export file or a running BigQuery dataset that outlives the 16-month window.

Step 7: Reconcile with your analytics

Search Console counts clicks from Google to your site. A web analytics tool counts sessions or events on your pages with its own rules. The two will not match, and that is expected. See GA4 Search Console integration for how to connect them and what each report means.

What success looks like: You use Search Console for search visibility and analytics for on-site behavior, and you do not add the two together.

Final verification checklist

The search type, date range and granularity are written next to every number you report.
You know whether each tab is aggregated by property or by page.
Dotted, preliminary data is excluded from conclusions.
Query-level conclusions allow for anonymized queries.
Average position is quoted only for a specific page and query set.
Exports are dated, and "~" and "-" cells are not mistaken for real zeros.
A BigQuery export or regular file export protects data older than 16 months.

Common mistakes to avoid

Treating average position as a ranking. It is an average of best positions across many queries. Filter to one page and one query before drawing a conclusion.
Expecting table rows to sum to the chart. Anonymized queries and property-versus-page aggregation explain most gaps. Compare like with like.
Judging CTR without position. A page at position 1 and a page at position 9 have different expected CTR. Compare similar positions.
Reading preliminary data as final. The newest hours can change. Wait for the line to become solid.
Waiting for the interface to keep everything. Rows older than 16 months disappear. Export on a schedule.
Adding Search Console clicks to analytics sessions. They measure different events. Report them side by side.

Frequently asked questions

Why do my impressions go up while clicks stay flat?

More people are seeing links to your site without clicking. That can come from new queries at low positions, a change in search features on the results page, or a less compelling snippet. Check the Queries and Search appearance tabs for what changed.

Why is the Queries total lower than the chart total?

Anonymized queries are excluded from the table but included in chart totals unless you filter by query. This is a documented privacy filter, not an error.

How far back does the data go?

Google's Analytics help says Search Console keeps data for the last 16 months. Use exports or BigQuery if you need longer history.

Can this report tell me why traffic dropped?

No. It shows what changed in clicks, impressions, CTR and position. Finding the cause needs more evidence, such as indexing, technical, and content checks.

Official references

Next step: See how Crawvix checks a website and capture a baseline once scanning opens. This guide provides technical information, not a guarantee of rankings, traffic or index inclusion.