How to read Google Search Console without misleading yourself

The four numbers on the performance report

Clicks and impressions are counts and behave sensibly. Position and click-through rate are averages, and averages of these particular things hide more than they show.

Clicks — people who chose your result

The number of times somebody clicked through to your site from Google. This is the only number here that maps directly onto visitors, and it is the one to trust. It counts the click rather than the visit, so it will not match analytics exactly and is not supposed to.

Impressions — times your result was there to be seen

A page appearing in results counts as an impression whether or not anybody scrolled far enough to see it. A result sitting at position 40 collects impressions from people who never looked past position 10. Rising impressions with flat clicks usually means you are appearing for more things rather than doing better at any of them.

Average position — the number most often misread

The average of every position every page held for every query in the period. A page at position 3 for its main term and 80 for fifty irrelevant ones has a terrible average and an excellent ranking. Average position moving is far more often a change in which queries you appear for than a change in how you rank for the ones that matter.

Average CTR — an average of an average

Clicks divided by impressions, across everything. Because it inherits the impressions problem, it falls whenever you start appearing for more marginal queries, which is usually a good thing. It is close to meaningless site-wide and genuinely useful filtered to a single query or page.

Use the filters, or the report will lie to you

Almost everything useful in the performance report comes from filtering, and almost every wrong conclusion comes from reading it unfiltered. The tabs beneath the graph — queries, pages, countries, devices — are the actual report; the graph is a summary of everything at once.

Two filters are worth applying as a habit. Filter to a single page and you can see which queries that page ranks for, which is the only way to know whether it is being found for what it is about. Filter to a single query and the average position becomes a real position rather than a blend.

The comparison mode is the other underused control. Comparing 28 days against the previous 28 is the fastest way to see what changed, and it is much more informative than looking at one period on its own.

Queries where you rank between about 8 and 20

These are the ones worth attention, because they are close enough to the first page that a modest improvement changes the traffic materially. A page at 12 is usually a page that is nearly right rather than one that needs rewriting, and it is a better use of an afternoon than a page at 60.

Pages with impressions and almost no clicks

Filter to a page, and if it is collecting impressions at a decent position and nobody clicks, the problem is usually the title and description rather than the content. That is a fifteen-minute fix on a page Google already likes, which makes it the highest-return work available in this report.

The indexing report, and the two statuses that sound the same

The pages report divides your site into indexed and not indexed, with reasons. Most of the reasons are self-explanatory and two are not, and the two that are not need completely different responses.

This is the single most valuable distinction in Search Console and the one most often missed, because the labels differ by one word.

Discovered — currently not indexed

Google knows the page exists and has not fetched it yet. Nothing has been judged. This is usually about crawl priority: Google has decided the page is not worth the request right now, which commonly means the site has more pages than its authority justifies, or the page is buried many clicks deep. Rewriting the page will not help, because nobody has read it.

Crawled — currently not indexed

Google fetched the page, read it, and chose not to include it. That is a judgement, and it usually means the page was thin, near-duplicate, or did not add anything to what was already indexed. This one is a content problem, and it is the one where rewriting is the right response.

Alternate page with proper canonical tag

Not a problem. It means the page declares another page as the main version and Google agreed. Seeing a large number here is normal on any site with filters, tags or pagination.

Duplicate, Google chose different canonical than user

This one is worth reading carefully. You told Google which version is authoritative and Google disagreed and picked another. It usually means the pages are more similar than you think, or your internal links point at a different version from the one your canonical names.

Sitemaps, and what submitting one does not do

A sitemap tells Google which pages you consider part of the site. It is a hint about discovery, not an instruction to index, and submitting one does not make anything rank. Its real value is diagnostic: the difference between the number of pages you submitted and the number indexed is one of the most useful figures in the whole tool.

A sitemap listing pages that redirect, return errors or carry a noindex is worse than a short one, because it spends crawl attention on addresses that lead nowhere and tells Google your list of your own pages is unreliable.

What Search Console will not tell you

It only reports Google. Bing, and increasingly the AI assistants that answer questions without sending a click, are invisible here. For most UK small businesses Google is the overwhelming majority of search traffic, so this is a reasonable simplification rather than a serious gap — but it is a simplification.

It holds 16 months of history and no more, which is why exporting before a migration is worth the five minutes. Query data is also sampled and thresholded: rare queries are omitted entirely to protect privacy, so the query list is never the complete picture and the totals will not add up.

And it reports what Google sees rather than what a visitor sees. A page can be indexed, ranking, and broken for everybody who arrives on it, and nothing in Search Console will say so. That is the gap a crawl of your own site fills, and it is the reason the two are worth using together rather than either alone.

The checks behind this guide

• Google indexes the pages it has looked atGoogle crawls far more pages than it indexes, and a page it has crawled and declined to index will never appear in a search result. It is not a penalty and there is usually no message about it — the page simply earns nothing. The common causes are pages that duplicate each other closely, pages with very little of their own content, and pages Google reached but judged not worth storing. Only Search Console knows this has happened; it is invisible from the outside.
• Google has crawled the pages it knows aboutGoogle keeps a list of pages it has found and works through it at a rate it decides. A page sitting on that list has never been fetched, so nothing about the page itself is the cause — it is not thin, it is not duplicated, it has not been judged at all. What holds the queue up is almost always the site as a whole: a server that responds slowly, far more URLs than there is real content, or pages nothing links to except the sitemap.
• Search terms close to the first page have been identifiedAlmost nobody goes to the second page of results, so a search term ranking eleventh earns close to nothing while a term ranking ninth earns steadily. Terms sitting just off the first page are the cheapest improvement available to most sites, because the work needed to move a page from eleventh to eighth is a fraction of what it took to get to eleventh in the first place.
• Your pages get clicked about as often as their ranking suggestsA page can rank well and still be ignored. What somebody sees before deciding whether to click is the title and the description Google shows, and a page whose click-through rate is far below what your own other results at the same position achieve is being passed over for a reason that has nothing to do with its ranking. This is the cheapest traffic there is to recover, because the page is already winning the hard part.
• No page has lost a significant share of its search trafficA page quietly losing the traffic it used to get is the hardest kind of problem to notice, because nothing breaks and no error appears anywhere. Comparing two equal periods against each other is the only way it shows up. The causes are usually external — a competitor moving above you, a change in what Google shows for that query — or a change you made and did not connect to the outcome.
• Google agrees with the canonical URL your pages declareWhen several URLs show the same content, a page can declare which one is the real one — the canonical. Google treats that as a strong hint and not as an instruction, and where it disagrees it indexes the URL it chose instead. That matters because everything the page earns is credited to Google's choice rather than yours: the URL you link to, share and measure may not be the one collecting the value.
• Google has fetched every sitemap it has been given, without errorsA sitemap is how a site tells Google which pages exist and when they changed. Search Console records what it has been given, when it last managed to fetch each one, and how many errors it found — and that record regularly disagrees with what the site is actually serving. A sitemap submitted years ago and never fetched since, or one full of URLs that no longer resolve, is worse than none: it is a source Google has learnt to distrust.
• Every page Google has indexed still loadsGoogle keeps showing a page in its results until it re-crawls and finds it gone, and that can take weeks. In the meantime every person who clicks that result lands on an error — the worst possible first impression, and one nobody inside the business ever sees, because they navigate to pages rather than searching for them.