Read the Search Console performance report as five shapes
The Search Console performance report gives four numbers and no diagnosis. Five recognisable shapes cover almost everything a page can be doing wrong, and each one has a different next action.
The Search Console performance report gives you four numbers and no diagnosis. Clicks, impressions, average position and click-through rate describe a symptom; they never name the illness. The useful skill is reading the four together as a shape, because five recognisable shapes cover almost everything a page can be doing wrong, and each one has a different next action.
Read this first
You need a verified property with at least three months of data. Below that, the shapes are indistinguishable from noise.
This chapter is about diagnosis. Acting on the most common finding — pages sitting just outside the first page — is covered in striking distance keywords, and the shape where two of your own pages compete is handled in the keyword cannibalization fix.
What the four numbers in the Search Console performance report mean
Two of them are more slippery than they look, and both mislead in the same direction — they flatter.
An impression does not mean anybody saw you. Google's definition: "an impression is counted whenever an item appears in the current page of results, whether or not the item is scrolled into view, as long as the user need not click to see more results". A result at the bottom of page one counts every time, viewed or not.
Average position is an average of your best placements, not of all of them. Google's documentation says the reported value is "the topmost position occupied by a link to your property or page in search results, averaged across all queries in which your property appeared", and that "a link must get an impression for its position to be recorded". Both facts read 2026-08-18 in the Search Console help.
An average position of 6 can mean you sit at 6 every time, or at 3 half the time and 40 the other half. The report shows one number for both.
Clicks are the one number here that means what it says. That is worth stating, because it is why the other three are worth the trouble: clicks tell you what happened but never how close you came, and the whole point of a diagnosis is to distinguish a page that nearly worked from one that never entered the running.
Click-through rate inherits both problems, since it divides real clicks by impressions that may never have been looked at. Treat it as directional between similar rows, and never as an absolute you can benchmark against somebody else's article.
The deliverable: five shapes and what each one means
Set the range to three months, open the Queries tab, and enable all four metrics. Then match each row against this table.
| What you see | What it means | Next action |
|---|---|---|
| Many impressions, almost no clicks, position 8 to 20 | You are on page two | Strengthen the page; this is a ranking problem |
| Many impressions, few clicks, position 1 to 5 | You rank but the result is not chosen | Rewrite title and description; check what sits above you |
| Few impressions, good position | The term is small, or you surface rarely | Check the term is worth the page at all |
| Impressions rising, clicks flat, position flat | Something above you changed | Search the term yourself and look at the page |
| Two of your pages on one term | They are competing | Merge, differentiate, or leave it deliberately |
Work top to bottom and stop at the first row that fits. The order is deliberate: the top two cover most pages, and the bottom row is the one people reach for first because it is the most interesting.
Shape two is the one worth slowing down on
Ranking in the top five and getting no clicks used to have one explanation, which was a weak title. Now it has three, and they need different responses.
The first is still the title. If your entry reads like a category label while everything around it answers a question, the fix is a rewrite and it is cheap.
The second is intent mismatch. You rank for a term whose searchers want something you do not offer — a tool, a price, a download — and they scroll past. No title fixes that, and the honest response is often to let the term go.
The third is that something is answering the question before your result. An AI summary, a featured snippet, or a set of expandable questions can satisfy a search without a click, and your impression still counts. You cannot see this in the report. You can see it by searching the term yourself, which takes ten seconds and is the step people skip.
How to run the whole pass
Once a month, half an hour.
- Open Performance, set three months, and sort the Queries tab by impressions.
- Take the top thirty rows and label each with one of the five shapes. Write the label down; the value is in having last month's labels to compare against.
- For every row you labelled shape two, run the search yourself and note what occupies the space above your result.
- Pick the three rows with the most impressions among shapes one and two. Those are your month's work.
- Leave everything else alone. A list of thirty problems is a list nobody works through.
Doing step two by hand is the part people want to skip, and it is where the value is. A label written down in March is what turns April's report from a snapshot into a comparison, and the report itself will not remember for you.
Step five is the one that makes this repeatable. The report will always show more problems than you have time for, and the discipline is choosing by impressions rather than by irritation.
Two readings that look like insight and are not
Both are common in reporting decks, and both survive because nobody in the room has the data to contradict them.
- Comparing your click-through rate to a published benchmark. Those tables average across industries, countries and result types. Your number is only meaningful against your own history and against your own other rows.
- Reacting to a single week. Position averages move on small samples, and a term with forty impressions can swing several places on the strength of one unusual day. Three months is the shortest window worth a decision.
Where this chapter stops
The Search Console performance report describes Google Search. It does not tell you what any answer engine did with your page, and it cannot tell you why a ranking moved — only that it did. Anyone reading causes out of these four numbers is inferring, and should say so.
It also shows your property in isolation. There is no competitor in this data, so a page that improved every month while a rival improved faster looks like unambiguous progress here.
Two more limits. Not every search term you appear for is listed, so the rows are a sample and not the full picture. And a page can be diagnosed correctly and still be beyond saving, when the result type has changed to something your page format cannot occupy.
Running this labelling every month without a human doing it by hand is what QueryWin is being built to do.
Common questions
Why does the Search Console performance report show a different position than I see?
Google's help says directly that "position value is the average position for all searches", and that your own result will differ because of history, location and other variables. Your personal search is one data point against an average.
Do impressions mean someone saw my listing?
No. An impression is counted when the result appears in the current page of results, whether or not it was scrolled into view. A listing at the bottom of the page counts the same as one at the top.
What click-through rate should I aim for?
There is no number worth quoting, and the published tables are averaged across contexts too different from yours to apply. Compare a row to its own history instead.
How often should I run this?
Monthly, with a three-month window each time. More often and you are reading noise; less often and you find out about a decline one quarter late.
Part of the QueryWin handbook · Level 2
