Analytics · 7 min read
How to read Search Console without fooling yourself
Search Console is happy to give you a number that looks like progress and means nothing. The fix is to stop reading averages and start reading distributions.
Every month someone sends me a screenshot of a site with “average position improved from 18 to 16” circled in red. Sometimes that is real progress. Often it means one new page started ranking for a long-tail phrase in position 40, and the average moved for reasons nobody can act on.
Average position is a weighted average of where you appear, across every query, device and location. It is a summary, not a diagnosis. Here is how I read the report instead.
1. Separate pages from queries
The Performance report has two tabs people treat as one. They answer different questions:
- Queries tell you what demand looks like. Use this to decide what content to create or rewrite.
- Pages tell you which URLs are earning. Use this to decide what to fix first.
My routine: start on Pages, find the pages with high impressions and low CTR, then switch to Queries filtered by that page to see which specific searches are letting it down.
2. Treat impressions as a ceiling, not a win
Impressions mean you appeared — not that anyone saw you. A page ranking in position 12 for a high-volume term can generate hundreds of thousands of impressions with almost no clicks, because most people never scroll that far.
So before celebrating impressions, ask: where is this page actually positioned? Impressions at position 3 are an opportunity. Impressions at position 30 are a report line.
3. Use position bands, not the average
Export the Queries tab and group positions into bands: 1–3, 4–10, 11–20, 21–50. Now you can see where the site actually lives, and which band has the most impressions stuck in it.
The 11–20 band is where most of the useful work is. Those pages are close enough that a title rewrite, an internal link or a content refresh can move them onto page one — and far enough that the improvement is visible within weeks.
4. Filter out brand queries early
People searching your company name will click. They inflate CTR, they inflate position, and they tell you nothing about acquisition. Create a filter that excludes your brand terms before you draw any conclusions.
I keep two views saved: one with brand, one without. If non-brand clicks are flat while total clicks grow, the growth is coming from people who already knew you.
5. Compare like with like
Search Console defaults to a three-month comparison, which is fine for a trend and terrible for a launch. If you published a page two weeks ago, compare the last 28 days with the 28 before it — and expect noise. If you changed a title yesterday, wait. It can take days or weeks for Google to recrawl and re-evaluate.
Also check device and country splits before you panic. A drop that only exists on mobile is usually a layout or speed issue, not an algorithm event.
6. Look for the pages that lost, not only the ones that gained
The Pages tab sorted by click difference, with the comparison set to the previous period, is the most useful view in the whole tool. It surfaces pages that quietly lost 40% of their clicks while the site total stayed flat.
Common causes: a competitor published something better, a title changed by accident, a page got merged or redirected, or the query itself became seasonal. Each has a different fix, and none of them is “write more blog posts”.
A monthly routine that takes 30 minutes
- Open Performance, last 28 days vs previous period, brand excluded.
- Check Pages sorted by click difference. Note the three biggest losses.
- Check Queries in the 11–20 position band with meaningful impressions.
- For the top five, compare CTR with expected CTR for that position.
- Write one action per item: rewrite title, add internal link, refresh section, merge page.
- Record the date and the change in a sheet, so next month you can tell what worked.
What this looks like in practice
On one site, the quarterly view showed 8.2% CTR and an average position of 13.1. The average hid two different stories: a handful of service queries in positions 4–8 doing most of the clicking, and a long tail of informational queries in positions 15–30 generating impressions and nothing else. Fixing the titles on the first group was a week of work. Trying to fix the second group by rewriting everything would have been a month of waste.
The report is not the strategy. It is the evidence you use to choose one.
Want a second opinion on your Search Console data? Request a free audit and I will send back the three things I would fix first.