Google Search Console is the closest thing most teams have to first-party search data. Its Performance report shows four metrics: clicks, impressions, CTR and average position. They look self-explanatory, which is exactly why they get misread in monthly reports.
This article walks through how Google defines each metric, the traps that follow from those definitions, and a simple way to read them together.
The four metrics, as Google defines them
Google documents its definitions in the Search Console Help page on performance metrics. The short version:
| Metric | Google’s definition (summarized) | What it is good for |
|---|---|---|
| Clicks | A click that sends the user to a page outside Google Search, Discover or News | Real visits generated from search |
| Impressions | A user has seen, or potentially seen, a link to your site | Visibility and demand |
| CTR | Clicks divided by impressions | How attractive your result is for the visibility you have |
| Average position | The topmost position of your link per query, averaged across queries | Ranking trend, with caveats |
Each of these has a nuance worth knowing.
Clicks
Only clicks that leave Google count. According to the same Help page, clicks that keep the user inside Google, and clicks on query refinement links, are not counted. Clicks are also not sessions: your analytics tool will rarely show the same number, because it measures something else (and depends on consent, tagging and redirects).
Impressions
An impression is counted, as a general rule, whenever your item appears on the current page of results. The user does not need to scroll to it on a standard results page. In carousels, expandable widgets and infinite scroll surfaces, the item usually has to be scrolled into view or expanded to register (source).
The practical consequence: impressions measure potential visibility, not attention. A rise in impressions often means you started appearing for more queries, sometimes far down the page.
CTR
CTR is a ratio, so it moves whenever either side moves. If you suddenly gain impressions for a broad new query at a low position, CTR can drop even though nothing got worse. Always read CTR next to impressions and position, never alone.
Average position
This is the most misunderstood metric. Per Google’s definition:
- For each query, Search Console takes the topmost position occupied by a link to your property (or page).
- It then averages those values across all queries where you appeared. Google’s own example: positions 2 and 3 on two queries give an average of 2.5.
- A position is only recorded when the link received an impression.
- Position is calculated for Google Search results only.
So “average position 8.4” for a whole site is an average of many very different queries. It is not where a typical visitor finds you.
Five common misreadings
1. “Our average position dropped, so rankings got worse”
Not necessarily. If your site starts appearing for hundreds of new long-tail queries at positions 40 to 60, the site-wide average falls while every existing ranking stays the same. Check whether impressions and the number of ranking queries rose at the same time.
2. “The chart and the table disagree, so the data is broken”
Totals in the chart and the table can differ. Google explains in the Performance report Help page that this is usually due to aggregation: the chart aggregates by property, while the table aggregates by page on the Pages and Search appearance tabs. When one result shows several of your URLs, grouping by property counts one impression, grouping by page counts one per URL.
3. “The latest days look bad”
The most recent data can be preliminary and may change in the next few hours. Avoid commenting on the last day or two in a client report.
4. “CTR is low, so the title is bad”
Sometimes. But a low CTR at position 9 is expected behavior compared with the same page at position 2, and some queries are answered directly on the results page. Compare CTR for pages with similar positions and similar query intent before rewriting anything.
5. “Search Console says the page has data, but the URL is different”
Data for page variants (for example separate mobile and desktop URLs) is generally assigned to the canonical URL. If your reporting joins Search Console data to a list of URLs, normalize on the canonical first.
How to read the four metrics together
A useful habit is to look at combinations rather than single metrics. A few patterns:
| Pattern | Likely reading | Next step |
|---|---|---|
| Impressions up, clicks flat, position down | New visibility on broader or deeper queries | Check which queries are new, not a ranking loss |
| Impressions stable, clicks down, position stable | Fewer clicks for the same visibility | Inspect the results page: new features, a rewritten title, competitors’ snippets |
| Impressions down, position stable | Less demand (seasonality) or fewer queries | Compare with the same period last year |
| Position up, CTR down | Moved up on low-intent or very visible queries | Segment by query |
Segmenting is what makes these readings reliable. The Performance report lets you break data down by queries, pages, countries, devices, search appearance and dates, and filter by search type (web, image, video, news) (source). A site-wide number mixes brand and non-brand queries, mobile and desktop, and every country you serve. Splitting at least brand versus non-brand and by country removes most false alarms.
A short checklist for monthly reporting
- Compare like with like: same number of days, and year over year for seasonal businesses.
- Separate brand and non-brand queries.
- Report average position per query group or per tracked keyword, not just site-wide.
- Read CTR next to position, never in isolation.
- Leave out the most recent, preliminary days.
- Note that exported values shown as “~” or “-” in the interface are exported as zeros, and treat them accordingly in spreadsheets.
- Annotate known events: releases, migrations, content launches.
How SearchFirefly helps
SearchFirefly brings in clicks, impressions, positions and pages from Search Console and shows them next to your tracked keywords, including movements in and out of the Top 3, Top 10 and Top 20. With multi-market views, trends are read per country or language instead of as one blended average, which removes many of the misreadings described above.