Search Console average position: report it or drop it from your reports?

Average position shows up in almost every SEO report, and hardly anyone can explain what it means. It's an average across every query a page appeared for, weighted by impressions. A page that ranks 1st for a branded query and 40th for thousands of generic ones can land around 12, which tells you nothing useful. It also drifts when Google adds new queries or tests a new layout, even if your rankings haven't moved at all.

I'd stop using it as a headline KPI. Report clicks and impressions by query group or by page instead. If you want a position signal, track a fixed set of 20 to 30 target keywords with a rank tracker that checks the same SERP on the same schedule. An average can show a rough direction, but it's a poor basis for a call like "the SEO work isn't working." Does anyone still put it in client reports, and how do you explain it?

If clients still insist on average position, only quote it for a fixed query set, and always next to impressions. The number moves on query mix alone. If a page starts picking up a flood of long-tail impressions at position 30, the average can drop while nothing improved for the terms that matter, and the client will read that as a win. If you can't name the queries behind a change, leave it out of the report.

Before a metric goes in a report, ask whether a move in it would change what the client does next week. If not, leave it out, even if you attach a query list. Average position mostly reflects where Google happened to show your pages for the impressions it gave you, which isn't the same as how well the SEO work is going. If nobody can say what action a rise or fall would trigger, the number is decoration and mostly just starts arguments. Clients usually respond better to something like "these five money terms moved from page two to page one, and that's what we're chasing." That gives them a decision to make, so the number doesn't have to carry the whole story.

Search Console doesn't average your rankings evenly. It records your best position on each impression and weights by impression count per query. So a page can sit at 30 for a hundred long-tail queries and at 3 for one money term, and the average still looks bad even though the money term is fine. The practical fix is to pull those five money queries into their own filtered view.

Also check whether your average shifted in late 2025. Google stopped honoring the num=100 parameter around then, which cut deep-result impressions for a lot of sites with no real ranking change behind it.

Check the num=100 change from the other direction too. The impressions that disappeared were mostly deep results, sitting at positions 30 to 100, so losing them tends to make your average position look better, not worse. If your average improved around late 2025 while clicks stayed flat, that's the likely cause. You can confirm it by comparing impressions by position bucket before and after the change. If the drop is concentrated in the 30+ buckets, that's your answer.

I don't think the confirmation step holds up. The Search Console UI doesn't break impressions out by position bucket, and the API only returns one average position per query row, so you can't cleanly isolate the 30+ results. A better check is to take a fixed set of queries and compare their impressions before and after the change.

Fair, the bucket check won't work with what the API returns. Whether a fixed-query comparison holds up depends on how much of your traffic comes from queries that show up on both sides of the change. If most of your impressions are long-tail terms that appear and vanish, the fixed list will miss most of the movement.

Jun's right that a fixed list misses the long tail, but for the num=100 question I'd compare clicks on the fixed set rather than impressions. The impressions that disappeared were mostly deep results at positions 30 to 100, and those rarely produced clicks anyway. So if the fixed queries' clicks held steady while their impressions fell, that points to a reporting artifact. If their clicks dropped, something real changed, and a tail explanation doesn't cover it.

How much this tells you depends on how much of your click volume comes from queries that existed on both sides of the change. If most of your clicks come from long-tail terms, the fixed list only shows a small slice, and you're back to reading totals.