What Search Console Cannot Tell You

Search Console is the closest thing to ground truth you get for free. It reports what Google recorded, on your property, without a browser in the middle dropping referrers. It is also, in four specific ways, quieter than people assume — and each silence has produced its share of confident wrong conclusions.

It does not report every query

A substantial share of your impressions are attached to queries Search Console will not name. Google filters out terms searched by too few people, because a rare query can identify an individual. The totals include them; the query table does not.

The practical consequence is that summing the query rows gives a smaller number than the reported total, and the gap can be large on sites serving specific niches. Anyone building a report by exporting the query table and summing it is under-counting, invisibly.

Average position is an average of averages

The position figure is the mean of the best position your page held each time it appeared, across every impression in the period. It is not where you rank, and it moves for reasons that have nothing to do with ranking changes.

A page ranking third for its main term and fortieth for a hundred incidental terms shows an average dragged toward forty. Start ranking for more incidental terms and the average worsens while the business improves. This is the single most misread number in the interface.

Use it only for a query you have filtered to, on a page you have filtered to, compared to itself over time. In that form it is meaningful. Site-wide average position is close to noise.

It only knows Google

Every number is Google's. Other search engines, AI assistants that do not route through Google, and internal site search are all absent. On sites with meaningful visibility elsewhere, Search Console describes a slice and says nothing about the rest.

This has become more relevant as answer engines route traffic. If an assistant cites your page and someone clicks through, whether that appears in Search Console depends entirely on whether the assistant is drawing from Google's results. Absence there is not evidence that AI-mediated discovery is sending you nobody.

It cannot tell you why a page was excluded

The coverage reports name a state — crawled and not indexed, discovered and not crawled, duplicate without a user-selected canonical — and stop. They do not tell you the reason, and the state names are easily over-read.

"Crawled — currently not indexed" is the clearest example. It has no single cause. It can mean the content is thin, that it duplicates another page, that it is not considered worth the index space, or that Google has simply not got to it. The report does not distinguish, and treating it as a verdict on quality is a guess wearing an official label.

Worse, the reports lag. A page fixed weeks ago can still sit in an exclusion bucket. The live URL inspection tool queries Google's current state and is the only way to know whether an exclusion is still true. When the report and the live test disagree, the live test is right and the report is stale.

What it is genuinely best at

Three things nothing else does as well. Query-level demand on your own pages, which beats every third-party volume estimate. Click-through rate by query and page, which tells you where a better title would earn more from rankings you already hold. And index state, which is the only place Google reports what it did with your URLs.

Those are exactly the questions worth taking to it. The trouble starts when it is asked questions it was never measuring — how much traffic you get, why a page underperforms, or how you are doing overall.

Check one thing today: filter the performance report to a page you care about, sort by impressions, and look for queries where you sit between positions five and fifteen. That list is demand you have already earned and are not yet collecting.