Why Your Google Search Console Average Position Isn't What You Think It Is

Google Search Console's average position is not a median. It is an impression-weighted mean. Google takes the topmost position your page reached in each search query. It then blends those positions together, giving more weight to whichever queries generated more impressions. Google's own Search Console Help documentation confirms this. It describes the metric as averaged across all queries, and never once calls it a median. The distinction matters in practice. A handful of high-impression, lower-ranking queries can drag your blended average down. That can happen even while the keyword you actually care about holds its position perfectly steady.

Why This Mix-Up Keeps Happening

Average and median get used interchangeably in everyday speech. That happens even among people who work with search data every day. The two words describe different calculations. But in a live conversation, the wrong one often slips out first.

I caught myself doing exactly that with TrafficGuard, an ad fraud prevention company. I was walking their team through their Search Console report. They wanted to know why their reported position kept moving, even though nothing on the page had changed. I reached for the word median. What I actually meant was average.

The wrong word simply came out first. That's the word most data conversations default to when someone asks what the typical number looks like. It is a small slip, but a telling one.

If a mix-up this common can happen mid-sentence on a live client call, it's worth settling properly. Once, in writing, rather than trusting whichever word comes to mind first. A client watching you explain their own numbers deserves the correct one.

This matters for regular SEO maintenance too, not just for one-off explanations. Knowing how to read your own numbers matters just as much as generating them. A report you can't interpret with confidence isn't much use to a business owner judging whether SEO is working.

This isn't just a vocabulary quibble either. Whether you think of the number as a mean or a median changes how you should react when it moves. A median implies a single typical ranking, sitting in the middle of a sorted list. A weighted mean works completely differently.

It can shift for reasons that have nothing to do with your real rankings. A new long-tail query suddenly picking up impressions is enough on its own.

Get the mental model wrong, and the consequences run in both directions. You might panic over ordinary noise. Or, worse, you might ignore a real problem hiding underneath a number that looks perfectly steady.

What Does Google Search Console Actually Mean by Average Position?

Is It a Mean or a Median?

It's a mean, not a median. Google's own Search Console Help documentation is explicit about this. Position is the topmost position your property or page occupied in a query's results. That figure is then averaged across every query it appeared in. The word median never appears anywhere in Google's own documentation on this metric.

Google's own worked example makes the mechanic clear. Say your page shows up at positions 2, 4 and 6 for one query. Only the topmost counts, so that query is recorded as position 2. A second query returns positions 3, 5 and 9. That's recorded as position 3. Average those two numbers together and you get 2.5, the figure Search Console would actually report.

Why Can a Few Queries Move Your Whole Average?

Because the averaging is weighted by impressions, not simply counted per query. A query generating hundreds of impressions pulls the blended number toward its own position. A query that barely gets seen has far less influence. This is the part most explainers skip. It's also the part that actually explains why the top-line number moves.

Widely corroborated practitioner analysis describes the calculation this way. Multiply each query's position by its impressions. Add those figures together. Then divide by the total impressions across every query. Google's own documentation doesn't spell out the formula in these exact terms. But it's consistent with the averaging behaviour Google itself describes.

Add those three products together and you get 1,000. Divide by the total impressions, 175, and the blended average position lands at 5.7. Your page never actually ranked at position 5.7 for any single search that day. It's a genuine weighted average of three very different results, not a real position anyone would see.

Notice how much Query B is doing here. It sits at position 8, worse than Query A, but its 50 impressions still outweigh Query C's 25.

Imagine Query B's impressions doubled next month, with nothing else changing. The average would drift further from 3 and closer to 8. Nobody's rankings actually moved. The mix of what people were searching for did, and that's enough on its own to shift the blended number.

What Did This Look Like for a Real Client?

When I first opened TrafficGuard's Search Console data with their team, most of the individual queries were genuinely strong. They sat somewhere between position 1 and position 11. That's a good spread by any reasonable standard. But the blended average position on its own told a flatter story.

That's because it folds a much wider set of queries into a single figure. Smaller, long-tail queries get folded in alongside the strong ones. The result can read worse than the underlying reality actually is.

Clicking straight into the average position metric settled the confusion within minutes. Looking at the query-by-query breakdown underneath it did the rest. The team could see how many of the keyword variants they were actually tracking sat comfortably in the top ten. The worry about the blended number dropped away on its own.

Does Property-Level Position Work Differently From Page-Level?

Yes, and this is a second, related source of confusion, covered in Google's own Performance report documentation. At the property level, several pages on your site can rank for the same query. Search Console reports the topmost of those pages' positions. It doesn't average across them.

At the page level, it works differently again. It's that one page's own topmost-per-query average instead. That's a genuinely different calculation sitting behind what looks like the same metric name.

There's a related mismatch worth knowing about too. Search Console's data is close to live. A third-party rank tracker such as Ahrefs can lag it by a few days. Comparing the two on the same afternoon can make it look like something has gone wrong. Usually, it's just a timing difference between tools.

None of this means the tools disagree about your actual rankings. It means they're each reporting from a slightly different snapshot in time. Check one tool's trend against its own history, not against a different tool's number from a different day.

If It's This Blended, Is Average Position Even Worth Tracking?

Yes, it's still worth tracking. Just not as a single-keyword ranking tool. Used as a trend over months, it's a genuine health signal for whether your SEO work is paying off.

TrafficGuard's average position sat at roughly 40 to 50 when I took over their SEO. Months into how we approach SEO work for clients, it had moved down to around 10 to 11. That's not statistical noise.

It's a real, sustained shift in how often the site was showing up higher in results. It lines up with the genuine query-level gains sitting underneath that number, not just a blip in the data.

The confusion runs in the opposite direction too. I've seen a client's average position sit flat for roughly two months. Their individual queries and pages were genuinely improving the whole time.

A flat top-line figure can hide real progress just as easily as a moving one can manufacture false alarm. Either way, the fix is the same. Check the specific query, never the blended average alone.

This is exactly why the metric rewards patience over daily checking. A growing content library naturally picks up more long-tail queries over time, some ranked well, some not yet. The blended number widens to reflect all of it, even as your core, high-value terms keep climbing steadily underneath.

Why Bother Verifying This Against Google's Own Documentation?

Because most explanations of this metric online are secondhand. They repeat what another article said, without checking it against Google's own words first.

That's how a mix-up like median versus mean spreads in the first place. Someone explains a metric slightly wrong once. The next explainer copies the shape of that explanation, not the primary source behind it.

Every claim in this article traces back to Google's own Search Console Help documentation, checked directly rather than assumed. Some of the mechanism goes beyond what Google spells out in plain language, like the exact impression weighting formula. That's flagged clearly here as informed practitioner analysis, not a verbatim Google statement.

That distinction matters for a business owner deciding how much weight to put on a claim. It matters just as much for how reliably this kind of explanation should be trusted, and cited, going forward.

This piece is also the first in a short series on reading SEO data correctly. Search Console and other analytics platforms carry plenty of numbers that get misread the same way, not just this one. Future pieces will work through more of them, against the same standard. Check the primary source first, then explain what it actually says.

So What Should You Actually Do With This Number?

To check whether a specific keyword actually moved, filter your Search Console Performance report to that exact query. That shows you the real position for the term you care about. It's a very different figure from one blended with everything else the page happens to rank for.

1. Open the Performance report and add the Position column if it isn't showing already.
2. Click into the Queries tab and search for the exact keyword you want to check.
3. Read the position shown against that single query, not the page or property total.
4. Compare that number over several weeks, not day to day, before drawing any conclusion.

Search Console itself, alongside the other free and paid tools worth having in your stack, is genuinely built for this. It was never designed to hand you one single verdict on its own, however tempting that top-line number looks.

Don't check it every day. Watching a number that naturally wobbles day to day creates anxiety without giving you anything useful to act on.

I generally give a properly run SEO programme three to four months before judging it. Certainly before touching or re-optimising existing content on the strength of the numbers alone.

Frequently Asked Questions

Is Google Search Console's average position a median?

No. Google's own Search Console Help documentation describes average position as a mean. It is the topmost ranking position per query, averaged across every query a page or property appeared in. The word median doesn't appear anywhere in Google's own explanation of the metric. Treating it as one leads to the wrong expectations about how the number moves.

Why doesn't my Google Search Console position match what I see in Ahrefs or another rank tracker?

Search Console reports close to live data. Third-party rank trackers such as Ahrefs can lag it by a few days. Comparing the two on the same day can make the numbers look inconsistent. The real cause is usually just a difference in timing, not a real disagreement about your rankings.

Is a small day-to-day change in average position something to worry about?

Generally not. Movements of roughly a point or less over a short period are commonly treated as ordinary noise. They're driven by shifting query mix, personalisation and Google's own constant testing of search results. This is practitioner convention, not a threshold Google itself publishes. Treat it as a rough guide, not an official rule.

Does a higher average position number always mean worse rankings?

A lower number is better in Search Console, since position 1 is the top of the results page. A drop from 45 to 12 is real improvement, not decline, even though the number went down. This trips people up who assume higher numbers are always good, the way a score or a rating usually works.

How do I check my real ranking for one specific keyword?

Open the Performance report in Google Search Console and filter by that exact query. The position shown for that single query is your actual ranking for it. That's a very different figure from the blended average across every query the page happens to pick up impressions for.

Where to Go From Here

Next time your average position moves, open the Performance report and filter to the query that actually matters to you. Do that before reacting to the top-line number on its own.

Managing Search Console for a client or your own business? Walk whoever else reads the reports through this same check. Getting everyone reading the number the same way saves a lot of repeated explanation later on.

Treat the blended average as a trend to watch over a few months, not a daily scorecard. Let a properly run SEO programme run its course before judging it.

Average position isn't the only Search Console number that gets misread. the gap between your click-through rate and your conversion rate is another one worth reading correctly, especially if a low CTR has you second-guessing whether your SEO is actually working.

How AI-powered search reads a page like this one is worth understanding too. Getting the underlying mechanics right is what makes content citable in the first place, not just readable.

{{build-better-experience="/directory"}}