Search Console logs every query a rank tracker fires as an impression of yours, and on one of our domains that was 86%

Last week I published that, with /goto making rank tracking expensive, Search Console is the reasonable plan B, and I measured how much a report moves when you switch sources. I was missing the small print, and it runs the other way: those same tools' crawlers also show up in Search Console.

Not as visits, which we already knew, but as impressions. When a tool queries Google to see where you rank, Google logs that query as an impression of yours. There was never a human. And on the domain where it shows up most clearly, our own rank tracking tool, that is 86% of everything.

What happened on mencoro.com between July and September

Mencoro is our tool and it launched recently, so it has the traffic you would expect: little. That is exactly what makes it a good laboratory, because anything odd stands out with no background noise around it.

In the 90 days to 18 September, Search Console reports 84,401 impressions and 136 clicks. A 0.16% CTR and an average position of 36.6. With those two numbers in a report, anyone would conclude there is demand and a ranking problem. Both are contaminated.

The daily curve explains itself.

Demand multiplied by 63 and clicks went down. No ranking story explains that. And one detail settles it: before, weekends did 21% fewer impressions than a weekday, which is how people behave. During the sweep, weekends do 13% more. Crawlers do not care that it is Sunday.

Where those impressions land

They are not spread out. Four pages take 72,968 impressions, 86.5% of the total, and between them zero clicks in 90 days.

They are exactly the four pages describing what a rank tracker does. Which is to say: the ones other rank tracking tools watch to see what the competition is up to.

The country and device split completes the picture.

That 4% of the volume produces 62% of the clicks. That 4% is the human traffic; the rest is machinery.

The fingerprint that gives the tool away

Everything above is circumstantial. The proof is in a slip: there are 249 queries carrying a number and a colon in front, along the lines of 24903: track seo ranking or 9438: seo ranking tracker. They add up to 5,991 impressions and zero clicks.

Nobody types that into Google. It is a tool's internal keyword identifier leaking into the query it fires at Google. The tool is, quite literally, signing the crawl with its own case number.

You can look for that fingerprint in your account in a minute: query filter, regular expression, ^[0-9]{3,}: . Two warnings from experience, because Search Console's filter does not honour the pattern to the letter and over-returns:

• It ignores the space after the colon. On a certification client it surfaced 17020:2026 and 17025:2017, which are ISO standards and perfectly human searches.
• And it lets HTTP status codes through: on our own site, 5 of the 11 queries it returned were variants of 404: not found.

With those two discarded by hand, what is left is the real fingerprint.

How far it spreads: 14 domains checked

I ran the filter across 14 properties in the portfolio, from ecommerce to industry, media and services. The fingerprint shows up in three, and all three have something in common.

The trace on isocialweb.agency is minimal, and GetSTAT, the one showing up on gscsal.com, is another rank tracker. In the other eleven, zero. And it makes sense: tools crawl their own sector's vocabulary, so the fingerprint turns up on SEO sites and not on a camping shop.

That said, it does not follow that a camping shop is unaffected. When a tool crawls "tienda de campaña" for one of its clients, the query it fires is "tienda de campaña", with no number in front, and that impression is indistinguishable from a person's. What differs between mencoro and the rest of the portfolio is not that some have bots and others do not. It is that on some the bot left its signature on.

What this breaks in a report

Three things, and all three are figures that go out to clients every month.

Average position stops meaning anything. On mencoro it is 36.6 because it is averaging top-100 sweeps. Filtered to Spain it drops to 18.2. Same site, same week, eighteen places apart depending on whose impressions you believe.

CTR collapses without anyone having done anything wrong. A global 0.16% invites you to rewrite titles and metas. The CTR of the human slice is 2.40%. There was nothing to fix there.

And impression growth reads as success. This is the dangerous one. Multiplying impressions by 63 in four weeks makes a beautiful shape on a Search Console chart. If someone shows it without looking at the clicks, they are reporting a third party's crawler as if it were their own SEO.

My recommendation: carve out the human segment before you look at anything

There is no way to exclude those impressions in Search Console, so do not waste time hunting for the checkbox. What you can do is never look at the aggregate.

Define your project's human segment (the country and device where your market actually is) and use only that for position and CTR. On mencoro it is Spain on desktop. On a Spanish ecommerce it will be Spain on mobile. Clicks are trustworthy raw, because the bot does not click: they are the one metric that comes out of all this intact.

And if you run into a rise in impressions with no rise in clicks, before celebrating it or explaining it, look at the weekend curve. If Sunday matches Tuesday, it is not your market that grew.

I end up where I left off last week, with one caveat I did not have then. Search Console is the best plan B there is because it is the only one that measures real clicks. But its impressions are not demand: they are queries, whoever they come from.