No, not on its own. Average position is a mean calculated across every recorded impression and query for a page or property, which means it can shift for reasons that have nothing to do with actual ranking improvement, most commonly because the mix of queries triggering impressions changed. A page can see every one of its individual keyword rankings improve and still show a declining average position if new, lower-relevance queries entered the impression mix and dragged the mean down, or the reverse: a page can show an improving average position purely because it stopped showing up for weaker queries, with no actual ranking gain on the terms that matter.
How the metric is actually calculated
Google’s own Search Console Help documentation describes average position as the average of the best position a page or query achieved across the recorded impressions, aggregated over the selected date range. Critically, this is calculated per query, then averaged, and the set of queries contributing to that average isn’t fixed, it’s whatever set of queries actually generated an impression for that page during the period being measured.
This matters because the query mix generating impressions for any given page is rarely stable over time. New queries enter the mix as Google’s understanding of a page’s relevance to different terms evolves, seasonal or trending queries can spike temporarily, and content updates can shift which queries a page even shows up for at all. Every one of these changes affects average position independent of whether the page’s actual ranking for its established, important keywords moved at all.
Concrete scenario where the metric misleads
Consider a page that ranks position 3 for its two core target keywords consistently, and Google Search Console previously recorded impressions only for those two terms. If, over the following month, the page starts also generating impressions for a dozen loosely related long-tail queries where it ranks around position 25 to 40 (a natural consequence of Google’s query-understanding systems recognizing broader relevance as the page accumulates more signals), the average position metric will show a significant decline, even though the two core keywords the business actually cares about haven’t moved at all. A team watching only the average position number would reasonably conclude rankings got worse, when the accurate read is that the page’s overall visibility footprint through more secondary long-tail term impressions.
The reverse failure mode is just as real: a page could show an improving average position because it lost impressions for a set of weak, marginal, low-relevance queries where it was ranking poorly anyway, mechanically pulling the average up, while its actual position on the keywords that drive real traffic and revenue stayed flat or even declined slightly.
Why this makes average position specifically risky as a trend metric
The core problem is that average position conflates two genuinely different things: how well the page ranks for the queries it’s already established for, and which set of queries are contributing to the average at all. A metric meant to track “did rankings improve” needs to hold the query set constant, otherwise movement in the metric can’t be cleanly attributed to ranking change versus query-mix change. Average position, as GSC calculates and reports it, does not hold that query set constant by default, since it reflects whatever impressions actually occurred during the reporting window.
This is a particularly common trap in executive reporting, where average position gets pulled as a single headline trendline because it looks like a simple, intuitive summary number. Presenting it without the underlying query-mix context risks either false alarm (a real improvement effort appears to be failing) or false confidence (an actual ranking problem gets masked by query-mix drift).
What to use instead, or alongside it
The more reliable approach pairs query-segmented tracking with, rather than instead of, average position. This means tracking average position (or actual rank) for a fixed, defined set of priority keywords over time, either via GSC filtered to those specific queries or via a dedicated rank-tracking tool that monitors a consistent keyword list regardless of impression volume changes. Segmenting branded versus non-branded queries is also useful, since branded query volume and position behave very differently from non-branded terms and blending them into one average obscures both.
Property-wide or page-wide average position can still be useful as a coarse, secondary signal, especially for spotting large directional shifts worth investigating, but it should never be the sole metric used to conclude that an SEO effort succeeded or failed.
Practical implication
Don’t report average position alone as evidence of ranking improvement or decline, particularly to stakeholders who will take a single trendline at face value. Build tracking around a fixed, named set of priority keywords using either GSC’s query-filtered view or a dedicated rank tracker, so that the query set being measured doesn’t shift on its own between reporting periods. Where average position is included in reporting, pair it with a note on whether the underlying query mix changed materially over the same period, and use query-segmented (not blended) reporting whenever the audience is going to make a decision based on the number.
