Search Console is free, it is your own data, and almost nobody uses it beyond glancing at a graph. Two of its four headline numbers are routinely misread, in ways that lead to the wrong work.

Search Console gives you four headline numbers: impressions, clicks, click-through rate and average position. Two of them mean what you would guess. The other two do not, and the misreadings are consistent enough to be worth naming.

What does average position mean in Search Console?

Average position is the mean position your page appeared at across every impression in the period, not the position it holds today. A page that ranks third for a common query and thirtieth for a rare one has an average somewhere in between, and that average describes no real result. Treat it as a trend line to compare against itself over time, never as an answer to "where do we rank".

The consequence in practice is that "our average position improved from 14 to 11" can mean any of several different things: you genuinely moved up, or you lost impressions on a poorly ranking query, or you gained impressions on a well ranking one. Two of those three are not improvements at all.

This is why TrustCtrl always shows position as an average with its period attached and never renders it as a rank. A dashboard that says "you are number 3" is telling you something Search Console did not say, and the number it is based on cannot support the claim.

Why do the query numbers not add up to the totals in Search Console?

Because the query report returns only the most significant rows, not every query. Rare and personally identifying searches are withheld, so the sum of the listed queries is always lower than the site total. This is expected behaviour rather than a bug: use the date-based totals when you need a true figure, and treat the query list as the top of the distribution.

People discover this the hard way, usually while building a report. The clicks in the query table come to 4,000; the site total says 6,200; an afternoon disappears looking for the missing 2,200. They are not missing — they are the long tail, spread across thousands of queries too rare to list individually.

The practical rule is to keep the two apart. Totals answer "how are we doing". Query rows answer "what are people asking". Adding query rows up to produce a total produces a number that is confidently wrong, which is worse than no number.

What is a good click-through rate in Google Search?

It depends almost entirely on position, so there is no single good number. The first result takes a large share of clicks and the share falls steeply below it, which means a 3% click-through rate can be excellent at position eight and poor at position two. The useful comparison is against other pages of yours at similar positions — an outlier there points at a title or description doing the wrong job.

Two other things suppress click-through in ways that have nothing to do with your writing, and both are worth ruling out before you rewrite anything:

  • The answer is already on the results page. Opening hours, a phone number, a conversion — if the searcher gets what they needed from the snippet, the impression is real and the click never comes. That is not a failure, it is a query with no click in it.
  • An AI overview or featured snippet sits above you. Increasingly common, and it changes what a given position is worth. It also changes the goal: on those queries, being cited in the answer matters more than ranking below it — which is the subject of getting cited by AI search.

How do I find quick SEO wins in Search Console?

Look for pages with many impressions and a click-through rate below what their position would predict — the ranking work is already done and the two lines of text in the results are failing. Then look for queries where you sit just past the first page, where a modest improvement changes the traffic materially. TrustCtrl reads your Search Console data and surfaces both patterns as a ranked list of what to fix first.

Both patterns share a property that makes them the best work available: they need no new content, no links and no development project. A page with strong impressions and weak clicks needs a better title and description — an afternoon's writing against demand that already exists.

We treat these as opportunities rather than as faults, and keep them in their own list separate from technical findings. A missing security header is a defect with a right answer. "This page could earn more clicks" is a judgement call about where to spend an afternoon, and filing the two in the same place would make the defects harder to take seriously.

Connecting it to the rest of the picture

Search Console tells you what happened in search results. It cannot tell you why, because it never sees your pages — it has no view of a title that got truncated, a page that only renders with JavaScript, or a sitemap listing URLs that no longer exist.

That is the gap TrustCtrl closes. It crawls your site in a real browser and reads your Search Console data, so a page with collapsing impressions can be put next to what changed on that page. Connecting the property takes a couple of minutes, we ask only for read access, and you choose which property maps to which website rather than us guessing.

Two practical notes. Search Console data lags by a few days, so today's numbers are not there yet and a flat line for the last couple of days is normal rather than alarming. And when a figure is genuinely unavailable, we say so instead of drawing a zero — an absent number and a measured zero mean very different things, and a chart that conflates them will send you chasing a drop that never happened.

If you want the commercial version of this data — what the organic traffic is actually worth — that is what SEO is worth in money.