Top queries, clicks, impressions, CTR, position: how to read the Search Console export
You exported your Search Console performance data, opened the CSV, and the five columns raise more questions than they answer.
Short answer
Those five columns are one row per query: the phrase someone typed, how many times they clicked you, how many times you appeared, clicks divided by impressions, and the average of the positions you appeared at. Two of them mislead by design. Impressions count an appearance at position 80 the same as position 2, and position is an average that hides its own range.
Top queries
The phrase typed into Google, not a keyword you chose. Search Console shows a maximum of 1,000 rows per export and anonymises rare queries, so the file is a sample of your demand rather than all of it. If your totals do not add up to the chart above them, that gap is the anonymised tail, not an error.
Clicks
Someone appeared, saw you, and came. This is the only column in the file that corresponds to a human arriving on your site, and it is the one to judge work by. Everything else in the export is a leading indicator of this.
Impressions, the most flattering number you own
You appeared in a results page. That is all it means. An impression at position 80, on page eight, where nobody scrolls, counts exactly the same as an impression at position 2. This is why a site can triple impressions while clicks stay flat: it is appearing more often, in places nobody looks.
CTR, and why it falls when things go right
Clicks divided by impressions, for that row. The trap is reading it site-wide. As a site starts ranking for more queries, most of them weak ones, impressions grow faster than clicks and the rate falls. On this dataset CTR fell from 29.25% to 2.45% while monthly clicks rose from 31 to 32,887. A falling CTR with rising clicks is growth, not decay.
Position, which is an average of averages
The mean of the highest position you held each time you appeared. A page showing 11.5 might be at 4 for one query and 30 for another, or at 4 on desktop and 20 on mobile. Never act on a position number without splitting it by query, device and country first.
Split brand from everything else before you read anything
Queries containing your name behave nothing like the rest. In this dataset brand queries ran at 31.55% CTR against 1.94% for everything else, and they were 49.3% of all clicks. Leave them in and every average in the file describes two unrelated populations at once.
The three cuts worth making immediately
First, non-brand rows at position 4 to 20 sorted by impressions: that is the work with the shortest path to clicks. Second, rows with high impressions and no clicks, which tells you whether you have a position problem or a snippet problem. Third, the same export a quarter apart, because search data lags the work that caused it by roughly three months.
What the file cannot tell you
Everything after the click. Whether the visitor found what they came for, whether they completed anything, whether the page was broken on their phone. On this site 37.16% of sessions contained a click on something that was not clickable, and no column in this export would ever show it.
Questions
- Why do the totals in my export not match the chart?
- Search Console caps the export at 1,000 rows and anonymises queries searched by very few people. The chart counts everything; the rows you can see do not.
- Is a 2% CTR good?
- It depends entirely on position and on whether the query is branded. At position 8 to 10 this dataset measured 0.81% on non-brand queries, so 2% there is strong. At position 2 it would be poor.
- Why is my average position a decimal?
- It is a mean across every impression in the period. A decimal does not mean you sat in one spot; it means the positions you held averaged out there.
Measured, not asserted
Free tool for this: Search Console Analyzer. No account, nothing uploaded.
Where this goes deeper
Every number on this page comes from one complete dataset: one product taken from zero to 250K+ users on search alone, with nothing spent on advertising. The full argument is chapter-20-averages-describe-nobody of the book. Five chapters are free to read.