Google Search Console guide: the reports that matter and what each number measures

Google Search Console is the only report that shows what Google itself recorded about your site: the queries you appeared for, the clicks and impressions each page earned, which URLs are indexed and which are not, and why. It is free, it needs verified ownership of the site, and five reports carry most of the value: Performance, Page indexing, URL Inspection, Sitemaps and Core Web Vitals. The quickest way to get value from it is to learn exactly what each number measures, because several of them are routinely read as something they are not.

This page is a tour of the reports that matter, what each one actually measures, and a monthly set of checks to run.

Setting up the right property

There are two property types, and the choice matters. A domain property covers every subdomain and both protocols, and it is verified through DNS. A URL prefix property covers one exact prefix, such as https://www.example.com/, and can be verified by HTML file, meta tag, Google Analytics or Google Tag Manager. Google's guide to adding a property covers both.

Add a domain property for the complete picture, and URL prefix properties for sections you want to read separately, such as a blog subfolder or a documentation subdomain. Data accumulates from the day of verification, so verify a new property before you need it. Give access through Users and permissions rather than sharing a login.

The Performance report, and what the four numbers mean

This is the report people spend most of their time in, and the one most often misread. Google's documentation for the Performance report defines each metric, and these are the four that need care:

Metric	What it actually measures	How it gets misread
Impressions	Your URL appeared in a results page the user viewed, whether or not it was scrolled into view	Treated as people seeing you, when a position 40 impression was almost certainly never looked at
Clicks	A click through to your site from the results	Compared directly to analytics sessions, which count differently and will never match
CTR	Clicks divided by impressions for the rows in view	Read as a quality score, when it mostly reflects average position and the result type
Average position	The average of the topmost position your site held, across the impressions in the selection	Read as a rank. It is an average across every query, device and location in the filter

Three behaviours to know before you report from this data. Totals across tabs do not add up, because a single impression can involve several queries and pages, so switching tabs changes the denominator. Some queries are withheld to protect privacy, so the query rows will sum to less than the totals. And the interface returns a capped number of rows, so use the export or the API when you need the full set.

Used properly, the Performance report answers questions a rank tracker cannot:

• Which queries bring clicks you did not plan for. Filter by page, sort by impressions, and read the queries. They tell you what the page is actually about to Google.
• Which pages sit just below the fold of the first page. Filter position between 8 and 20, sort by impressions. These are the cheapest wins on the site.
• Brand against non brand. Add a query filter excluding your brand name. Non brand is the number that shows the channel growing.

Page indexing: what Google kept and what it did not

The page indexing report splits your known URLs into indexed and not indexed, and gives a reason for each exclusion. The reasons are the useful part:

• Crawled, currently not indexed. Google saw the page and chose not to keep it. Usually a quality or duplication judgement, so improve or consolidate the page rather than resubmitting it.
• Discovered, currently not indexed. Google knows the URL and has not fetched it. Often a crawl budget or server speed question on large sites.
• Duplicate, Google chose a different canonical. Your canonical was overruled. Compare the two pages and either consolidate them or make them genuinely different.
• Excluded by noindex tag and blocked by robots.txt. Correct for admin and filter pages, and worth reading line by line, because this is where an accidental block shows up.
• Soft 404. The page returned 200 and looked empty or like an error to Google. Common on out of stock products and empty search results.

Read this report by trend, not by chasing every row to zero. A healthy site has plenty of deliberately unindexed URLs. What matters is that the indexed count tracks the number of pages you want indexed, and that no exclusion reason grows without an explanation.

URL Inspection: the single page diagnostic

The URL Inspection tool answers questions about one URL: is it indexed, when was it last crawled, which canonical did Google pick, what did the rendered HTML look like, and which structured data was detected. Two things to know. The default view shows the indexed version, which may be weeks old, and Test live URL fetches the page now, so compare the two when you have just changed something. And the rendered HTML in the live test is the fastest way to confirm that JavaScript content is visible to Google.

Use Request indexing for individual important pages after a change. For a large batch, submit an updated sitemap and let normal crawling do the work.

Sitemaps, Core Web Vitals and Crawl stats

The Sitemaps report shows what you submitted, when it was last read and how many URLs were discovered. Discovered is not indexed, so read it alongside the indexing report. Split large sitemaps by section, because a split sitemap tells you which part of the site has the problem.

The Core Web Vitals report uses field data from real Chrome users, grouped by similar pages, which is why it lags a fix by weeks and why a lab tool can show green while the report still shows amber. Fix the template, then wait for the twenty eight day window to roll forward. Google's reference on the metrics explains the thresholds.

The Crawl stats report, under Settings, shows requests by response code, file type and purpose. It is where a server problem or a redirect loop becomes obvious when indexing slows on a large site.

A monthly set of checks

• Performance, last three months against the previous three, non brand only, at site level and then by page group.
• Any page group down more than ten per cent, opened by query to see what it lost.
• Page indexing: the indexed count, and any exclusion reason that grew.
• Sitemaps: last read date, and discovered counts per sitemap.
• Core Web Vitals: URL groups that moved between buckets.
• Crawl stats: 5xx and 404 rates, and average response time.
• Links report, for new referring domains and for anything unexpected.
• Export the Performance data and keep it, because the window is rolling and old data is only available if you saved it.

This page covers the Search Console reports for Google Search. For Discover and News performance, those tabs appear when a site qualifies and are read the same way. For analytics on what visitors do after the click, that is your analytics platform, and the two datasets answer different questions.

Asked and answered: Google Search Console guide

Why do Search Console clicks not match analytics sessions?

They count different events. Search Console counts clicks on a result; analytics counts sessions that loaded your tracking code. Redirects, slow loads, consent choices and ad blockers all sit between the two. Use Search Console for what happened in the results and analytics for what happened on the site.

What does average position really tell me?

It tells you the average topmost position your site held across every impression in the current filter. It is useful as a trend for one query on one device, and it is too blended to act on at site level. For a single term, filter down to that query.

How far back does the data go?

Sixteen months in the interface. Export the data you want to keep, monthly, so your history keeps growing past that window.

How long does indexing take after I request it?

Anything from hours to a few weeks, depending on the site and the page. Request indexing for the handful of pages that matter, keep the sitemap current, and use the indexing report to watch the batch.

Why do my query rows add up to less than my total clicks?

Rare queries are withheld to protect user privacy, and the interface caps the number of rows it returns. Totals are complete; the query list is a subset of them.

Queries this report has no tab for

Performance shows the queries that produced impressions on Google. It cannot show queries put to an assistant, because those never touch a results page.

Use the report as a prompt list. Pick your top fifty queries, rewrite the question ones in full sentences, and ask them in an assistant. Where Search Console says you rank and the assistant names a rival, you have found a page to sharpen. AI Knows Us records who is named across ChatGPT, Gemini, Perplexity, Claude and AI Overviews, so the monthly check sits next to your Search Console review.