What Is Average Position in Search Console?

Average position is the mean of the highest position your site occupied each time it appeared, averaged across impressions. It is accurate at exactly that job and misleading at every other one. It is not a rank tracker, it does not describe where you sit on the page, and it moves for reasons that have nothing to do with your rankings.

How Google actually calculates the number

Three rules produce the figure, and every confusing thing about it follows from them.

Rule 1 — Only the topmost link counts. If your site appears twice for one query, at positions 4 and 12, Search Console records a 4. The second listing is invisible to the metric. A site with sitelinks, an image result and a blue link for the same query still contributes one number.

Rule 2 — It is averaged per query, per day, weighted by impressions. The number attached to a page is not that page's rank. It is the average of its positions across every query it appeared for, with high-impression queries pulling harder. A page that ranks 2nd for one term and 45th for thirty others reports something in the thirties.

Rule 3 — Position is ordinal, not spatial. Position 3 means third in the result set. It says nothing about how far down the screen you are. Four ads, an AI Overview and a "People also ask" box can sit above the third organic result, and the metric will still read 3.

Google documents the mechanics in the Search performance report help page. The mechanics are not in dispute. The interpretation is where sites go wrong.

Is average position accurate?

Yes — and it still shouldn't be used the way most people use it.

The metric is computed from Google's own serving logs. There is no sampling error in the position values themselves. What makes it unreliable as a ranking read is that it answers a different question than the one you're asking. You want to know "where do I rank for my target keyword." The metric answers "across every query where anyone saw me at all, weighted by how often, what was my average topmost placement."

	Search Console average position	A rank tracker
Measures	Placement where impressions actually occurred	Placement for a fixed keyword list
Location / device	Whatever real users had, all mixed	Pinned by you
Personalisation	Included	Suppressed
Zero-impression keywords	Absent entirely	Still reported
Covers	Every query, including ones you never targeted	Only what you told it to check

Neither is wrong. They are different instruments. Search Console tells you what happened; a rank tracker tells you where you stand on a controlled test.

Why it moves when your rankings didn't

This is the part that causes the most wasted analysis, and it's a genuine statistical artefact rather than a bug.

Your average position can improve because you lost rankings. Suppose you rank 5th for one query and 60th for two hundred long-tail queries. Your average sits somewhere miserable. Now Google stops showing you for those two hundred weak queries. You've lost visibility — and your average position leaps to 5. The chart looks like a triumph. It is a contraction.

Your average position can worsen because you gained rankings. Publish a strong page, start surfacing for fifty new queries at position 30-something, and the average drops even though you added traffic. New content almost always drags the sitewide number down for a few months. This is normal and should not be treated as a decline.

The rule to internalise: average position is a ratio, and both halves of it move. A change in the number tells you the mix of queries changed. It does not tell you which direction is good. Any time you see a sharp move, check impressions alongside it before concluding anything.

Position improved, impressions fell → you lost long-tail coverage
Position worsened, impressions rose → you gained long-tail coverage
Position improved, impressions rose → genuine progress
Position worsened, impressions fell → a real problem worth investigating

That four-way check takes ten seconds and prevents most of the false alarms. The same discipline applies to impressions and clicks, where the pairs matter more than either metric alone.

Four more reasons the number misleads

Search appearance types are mixed together. Position 1 in the Images tab and position 1 in web results are both recorded as 1. If a page picks up image or video traffic, its blended average will look better than its blue-link ranking. Use the Search appearance and Search type filters to separate them before drawing conclusions.

Anonymised queries are excluded. Google filters out rare queries to protect user privacy, so the query-level rows never sum to your totals. The averages you calculate by hand from the query table will not match the headline figure.

The default view blends countries and devices. A page ranking 4th in the UK and 40th in India reports the middle. Filter to one country and one device before comparing anything to a rank tracker, or you are comparing a blend to a pin.

AI Overviews are folded in, not broken out. Google has said it includes AI feature performance within overall Search performance data rather than reporting it separately, and its AI features guidance confirms there is no AI-specific Search Console reporting. So some of what your position column describes is placement inside an element you cannot isolate.

How to read it usefully

Stop looking at the sitewide average. It is a mix-of-queries indicator, not a performance indicator.

Filter to a single query. This is the only level at which the number means "our rank." Performance → Search results → add a Query filter set to exactly matching.
Pin the search type, country and device. Web results, your primary market, one device. Now the number is comparable week to week.
Exclude branded queries with a doesn't contain filter when assessing SEO performance, for the same reason CTR benchmarking requires it — brand terms sit at position 1 and flatter everything.
Compare 28 days to the previous 28 days, not day to day. Daily values swing on query mix alone.
Read position next to impressions, always. Never in isolation.

Used that way it becomes a decent early-warning system: a single query drifting from 3 to 8 over a month, on a pinned filter, is a real signal worth acting on.

What to do when a specific query really is slipping

Once you've confirmed the decline is genuine rather than a mix effect, the causes fall into a short list: a competitor published something better, your page went stale, the intent behind the query shifted, or someone above you gained authority you didn't.

The first three are content problems — check content decay and refresh accordingly. The fourth is the one content can't fix. When the pages above you are comparably good and still outrank you, the difference is usually the strength of the link profile behind them, and closing that gap is slow, manual work. It is the problem Backlinkster was built for: one-for-one in-content link swaps between real site owners, verified live by code rather than taken on trust. Five swaps a month on the free tier, with paid plans from $19.

Frequently asked questions

What is a good average position in Search Console? There isn't one, because the figure depends entirely on how many low-ranking queries you appear for. A site with broad long-tail coverage can average 30 and outperform a site averaging 8. Judge individual queries under a fixed filter instead of chasing the sitewide number.

Why is my average position 8.4? There's no such rank. Because it's an average across impressions, not a rank. A page appearing at position 3 for some queries and 15 for others reports the weighted middle — a value that describes no single search result.

Is average position the same as my Google ranking? No. Your ranking is your placement for one query, in one location, on one device. Average position blends every query, country, device and search appearance where an impression occurred, and only counts your topmost link each time.

Why did my average position improve while traffic dropped? You most likely stopped appearing for a batch of low-ranking queries. Losing page-four impressions removes the worst values from the average, so the number rises while actual visibility falls. Check whether impressions fell at the same time.

Does average position include AI Overviews and featured snippets? Yes. Placement within Search features is counted in the same position numbering, and Google does not report AI feature performance separately from overall Search performance in Search Console.

Why doesn't Search Console match my rank tracking tool? They measure different things. Search Console reports positions where real impressions happened, personalised and blended across locations and devices. A rank tracker queries a fixed keyword list from a pinned location with personalisation suppressed. A gap between them is expected, not an error.

How far behind is the data? Search Console performance data typically lags by two to three days, so the last couple of days in any chart are incomplete and will revise upward.

The bottom line

Average position is a precisely calculated number that answers a question almost nobody is actually asking. Read it sitewide and you will misdiagnose growth as decline and contraction as progress. Read it one query at a time, with the search type, country and device pinned, alongside impressions — and it turns into a genuinely useful early-warning system. The metric isn't inaccurate. The default view is just the wrong altitude.