Search Console is the account Google keeps on your site: the queries it showed you for, the clicks you got, the pages it indexed and the ones it left out. It is the only search traffic source that comes from the engine itself. Semrush, Ahrefs and rank trackers estimate; Search Console measures.
A number from the engine does not argue with a number from an estimator. When the two disagree, the estimator is wrong.
1. The problem it settles for me
The same month opens the same way. A founder writes that traffic is down, the analytics tool shows a minus something, and the conversation scatters: seasonality, a competitor, an algorithm update, the spring redesign.
I answer nothing before opening Search Console. Is the drop on impressions or on clicks? On one page or on a hundred? On brand queries or on everything else? Those three questions split into three unrelated diagnoses, and none of them can be settled in an audience tool, which sees sessions and never queries. That is why Search Console is the first thing I open in an SEO audit.
2. How I use it, and the number that changed my method
I work through the API rather than the report. The reason is written into Google's own page on Search Console data: report tables never exceed 1,000 rows, and the API returns far more. On a B2B catalogue I look after, the gap is not cosmetic.
Three figures pulled on the same day, from the same property, for 1 May to 31 July 2026:
| What I asked for | Impressions | Clicks |
|---|---|---|
| Site total, no breakdown | 1,029,364 | 13,116 |
| Sum of the 41,031 queries returned | 697,524 | 5,946 |
| Sum of the 52,459 URLs returned | 1,126,632 | 13,401 |
32% of that site's impressions and 55% of its clicks sit under no visible query at all. Google calls them anonymised queries: the ones that are too rare, or that carry personal data, are never stored.
The third row is larger than the first, and that is the part nobody expects. The site total counts one impression per results page; the URL breakdown counts one per address shown. Two pages from the same site on one query are worth a single impression in the total and two in the detail.
So I never present the sum of a table as a site total. In a monthly report both numbers appear, and they say two different things. A client who adds up the rows of an export will always find a gap against the curve, and they are right to find it.
3. What I never ask it for
A ranking. The column is labelled average position and the label misleads. Google states in the performance report help that the figure reported is the topmost position held by the site. That is an average of best ranks, not the rank a searcher would have seen. A page that lands third on some days and fortieth on others comes out around 21, a figure that describes no single day. When a client wants to know where they stand, I use a dedicated rank tracker.
Search volume. Search Console only knows the queries the site has already been shown for. A keyword you do not appear on is simply absent, which says nothing about how many people type it: that is DataForSEO's job here. The reverse is my safety rule, and it has saved me more than once: a volume reported as zero is never on its own a reason to delete a page. I check the impressions Search Console actually credits it with first.
A competitor. Nothing at all. Google returns your property and only yours, which is fair enough.
A long history. The report keeps 16 months. A three-year comparison is prepared in advance, it is never recovered afterwards. I export every month for every property I follow: one extra call, and it buys an archive.
Today's number. Data lands after two or three days, and the day is cut on California time. On a Friday release, Monday still tells you nothing.
4. What I plug in around it
DataForSEO for demand, since Search Console only knows what the site has already captured. Google Analytics 4 for what happens after the click, knowing the gap between the two is normal: Search Console counts clicks inside Google's results, an audience tool counts visits on a page that managed to load its script. They will never measure the same thing, and I set out why in what GA4 no longer sees. Screaming Frog and Sitebulb for site structure, which Search Console does not show: it reports what Google saw, never what the site holds.
On call volume, Google publishes its thresholds: 1,200 queries per minute per site on performance data, 2,000 URL inspections per day. A monthly report across a client portfolio never comes near the first; a page-by-page indexing check on a large catalogue hits the second before the afternoon is out.
Telling measurement from estimate
Search Console does not tell you what to do. It tells you what happened, and it does so with gaps that Google documents. The whole craft is knowing which ones. Read those tables without knowing their blind spots and you build recommendations on sand; know them and you finally have ground to stand on.
Want to know what Google actually measures on your site?
I go through your queries, your pages and your measurement gaps, and tell you what sits behind the curve. It is where my SEO and analytics engagements start.
Request a free audit