Google Search Console
Google's free property-verified reporting tool for how a site appears in Google Search — and the reason most teams have no equivalent record of how they appear in AI answers.
Karl-Gustav Kallasmaa, Founder & CEOLast updated Google Search Console is a free Google service that reports how a verified site appears in Google Search — which queries showed it, how often those impressions turned into clicks, which URLs Google indexed, and which technical problems blocked it. Google's own description is that it "helps you monitor, maintain, and troubleshoot your site's presence in Google Search results."
Access is per property, not per user: you claim a domain or a URL prefix, prove control of it, and see data only for what you proved. That is why Search Console shows your data and never a competitor's, and why every "competitor's Search Console data" is an inference from a third-party estimate rather than a reading.
What it actually reports
Google lists five capabilities: confirm that Google can find and crawl the site; fix indexing problems and request re-indexing of new or updated content; view Search traffic data — how often the site appears, which queries show it, how often searchers click through; receive alerts when Google encounters indexing, spam, or other issues; and show which sites link to the site.
The Performance report carries exactly four metrics — clicks, impressions, CTR, and average position — filterable by search type across web, image, video and news results. Average position is the position of the topmost result from your site: across the whole site in the chart, and for that URL or grouping in a table row. It is not a keyword rank tracker, and reading it as one produces confident nonsense whenever a page gains or loses a second listing.
The limits that matter when you quote a number
Three are documented and worth writing on the wall.
- The data is capped, and it is capped at the top. The Search Analytics API returns 1,000 rows by default and a maximum of 25,000, and Google states plainly that it "does not guarantee to return all data rows but rather top ones," sorted by clicks descending. A long tail exists below the cut and is invisible, so "we rank for 25,000 queries" is a ceiling artefact, not a finding.
- The newest data is preliminary. Google notes that recent data is still being collected and might change within hours. Whole days compared against whole days is the only safe comparison.
- Queries are filtered before you see them. Google withholds anonymized queries — those not issued by more than a few dozen users over a two-to-three month period — from the table while still counting them in chart totals, which is why the sum of query-level clicks does not reconcile with the site total. The gap is not a bug to chase.
Why an AI-visibility team still opens it
Because two of its reports are upstream of everything an assistant can do with your page.
An answer engine that cites you has to fetch you first. The indexing and crawl reports are the cheapest existing evidence that your URLs are reachable, canonical, server-rendered and not blocked — the same preconditions that decide whether a retrieval step can use the page at all. See crawling and indexing, robots.txt, and XML sitemaps.
What it cannot do is tell you what an assistant said. Search Console's surfaces are Google Search surfaces. When a model names your brand in an answer, or an agent fetches your documentation page to answer someone else's question, nothing appears in the Performance report. The record of that event lives in your server logs, in the assistant's own citation list, or in a tool that samples model answers directly — not here. Treating a flat Search Console line as evidence that AI referrals are not happening is a category error, and it is the single most common one on this topic.
Failure modes
Reading average position as a rank. It is a topmost-result average over impressions. A page that appears twice, or that gains a featured-snippet-style treatment on some queries and not others, moves this number for reasons unrelated to ranking better or worse.
Comparing an incomplete window. Today is partial by design. A morning check against last week's full day always looks like a decline.
Mistaking the row cap for the truth. Exports stop at the documented limit. Coverage conclusions drawn from a truncated list overstate concentration and understate the tail.
Using CTR as a quality score. Click-through rate is a ratio of two capped, differently-filtered metrics, and it falls whenever a query set widens into impressions that were never going to click. A falling CTR alongside rising impressions is usually reach, not decay.
Expecting it to cover AI surfaces. It reports Google Search. Anything an assistant does with your content is measured somewhere else or not at all.
Frequently asked questions
Is Search Console free?
Yes. Google describes it as a free service, and signup is not required for a site to appear in Search results.
How much data can I export?
1,000 rows per API request by default, 25,000 at most, and Google states the response is top rows rather than all rows.
Does it show competitor data?
No. Access is granted per verified property.
Does it record AI assistant citations?
No. Its search type filter covers web, image, video and news results in Google Search.
Terms related to Google Search Console
The two separate stages that decide whether a page can be retrieved at all, and the reason a serving rule on a blocked page is never read.
A machine-readable list of the URLs you want discovered, bounded by the protocol at 50,000 URLs and 50MB per file, and a hint rather than an instruction.
The standardised file that tells crawlers what they may fetch, now the main place where AI training and AI retrieval access are decided separately.
Clicks divided by impressions, as Google defines it in Search Console — and why the ratio stops describing your reach once answers are generated instead of listed.