(Measurement)

Average Position

The most misunderstood of the four search performance metrics

Average Position is one of the four numbers shown in the Google Search Console Performance report: clicks, impressions, click-through rate, and average position. It is the average of the spots that links to your site held in search results, across every query that generated an impression. Even if your page appears several times in one search result, only the highest one counts. GSC Help Center

It is called a position, but it behaves like an average. So the sentence “our site’s average position is 12.4” does not mean you are 12th on a leaderboard. It is a number that squeezes countless spots, recorded across different queries, different pages, and different regions, into a single cell. If you miss this, things can look bad when they are going well, or the other way around.

Two rules: only the top spot, and only what was actually shown

The calculation rules described in Google’s Help Center are surprisingly simple. But that simplicity is exactly what causes the illusions discussed below. GSC Help Center

  • One spot per query. To quote the official example: if your site appeared at positions 2, 4, and 6 for query 1, it counts as 2. If it appeared at 3, 5, and 9 for query 2, it counts as 3. The average position across the two queries is 2.5. GSC Help Center
  • If it wasn’t seen, it isn’t counted. A dash in the position column is not an error. It means users never saw your site for that query, so there is no recorded position at all. GSC Help Center
  • What counts as an impression (the denominator). Unless the spot is one that users can only see by clicking “more,” an item counts as an impression as soon as it is on the current results page, whether or not it was scrolled into view. However, spots that must be scrolled or expanded separately, such as carousels and expandable items, and infinitely continuing lists only count as impressions once they come into view. In other words, separately swiped carousels and click-to-expand results are exceptions, and count as impressions only when they come into view or are expanded. GSC Help Center
  • Some surfaces have no position at all. Discover and News do not record a position value. Traffic from those surfaces needs to be judged by other measures. Surfaces with a different search type, such as image search, do record a position, but count it differently. GSC Help Center

Why it keeps disagreeing with what you count on screen

Google counts positions from top to bottom in the main area of the page, then top to bottom in the secondary area. GSC Help Center Image search counts differently: left to right, then top to bottom. For right-to-left languages, the direction is reversed. GSC Help Center

The unit of counting matters even more. Each element in the search results takes up one position, no matter how many links it contains. GSC Help Center An AI Overview takes up one position as a whole, and every link inside it gets the same position. GSC Help Center A horizontally swiped carousel is likewise one position, and every item in it shares the same position. GSC Help Center This is why the number disagrees with counting blue links on screen with your finger. The more complex the search engine results page (SERP) becomes, the wider the gap.

It is also normal for it to differ from the position you see when you search yourself. Google’s documentation states plainly that variables such as search history and location can make the position for a specific search differ from the average. GSC Help Center Google’s John Mueller, answering the same question in 2021, cited personalization, geo-targeting, and cases where a site appeared only briefly in search results. Search Engine Journal He also made clear that this data is based not on theoretical rankings but on search results actually shown to users. In other words, it is an observed value. Search Engine Journal

What the average hides: the truth behind a good-looking number

This is the part that needs the most care. Because result types of very different natures are mixed into one metric, spots that hardly get any clicks can drag the number down, or inflate it.

Google’s official positionAverage position is the average of the topmost spot held by links to your site, across all queries. An AI Overview takes up one position in search results, and every link inside it is given the same position value. GSC Help Center
Expert viewBecause result types of different natures are combined into one metric, in the past low-click spots in the corners dragged positions down, and now the balance has tipped toward highly visible top spots that don’t get clicks inflating the number. When you are cited in an AI Overview, the links inside get the top position value, so even if your blue link is far lower for the same query, the recorded position is the single top one. The number looks good, but clicks may come from the lower spot, so the position and your actual experience diverge. Note that this view is an industry contributor’s interpretation, not measured data. Search Engine Journal

It is similar at the page level. Even if a page holds the No. 1 spot for its target keyword, the many other queries it also appears for are mixed into the average, so the number comes out much lower. SEOTesting

This creates an improvement paradox. If you improve your content well and the range of queries the page appears for widens, the average after the improvement can come out lower than before. SEOTesting Concluding that your ranking dropped just because the average went down is a misjudgment.

That is why some even advise ignoring the site-wide average position in most cases, since it tells you almost nothing. Instead, they recommend scrolling down the Performance report to look at the average for each query. Practical Ecommerce An older warning not to treat this metric as a simple number makes the same point. A site splits into many pages, a page into many queries, and a query into many countries, and each has a different position. Analytics Edge

How to use it in practice: slice it finely and doubt your filters

It is not a useless metric. Once you drill down to individual queries, or to the list of queries a specific page appears for, average position becomes a basis for decisions. SEOTesting In other words, trust it only at the finest level of detail.

  • Look at positions 3 to 8 first. Ahrefs recommends filtering for an average position below 8.1 and sorting by position in descending order to pick out queries where a small push would bring a big traffic gain. Then click the query and check the Pages tab to see which page appears. Ahrefs data
  • Judge value by converting it to click-through rate. Moving from position 8 (1.8%) to position 2 (13.94%) for a query searched 1,000 times a month takes you from 18 to 160 clicks a month, about 9 times more. Ahrefs data Make a habit of reading the value of one or two positions on the click-through rate curve.
  • Don’t compare data before and after filtering by the same yardstick. Applying a page filter or a search appearance filter changes how data is aggregated, so totals can even go up after you add a filter. GSC Help Center
  • Some queries never appear in the table. The table shows at most 1,000 rows, so rare or long-tail queries drop out of the table and are counted only in the chart totals. GSC Help Center On the Queries tab, anonymized rare queries are also left out of the table. GSC Help Center This is also why the “matches” and “doesn’t match” filters don’t add up to the total. GSC Help Center
  • If chart totals and table totals differ, check the aggregation unit. It is usually the difference between site-level and page-level aggregation. GSC Help Center The Search Console API documentation also defines position simply as average position, then leaves the explanation of how site-level and page-level calculations differ to the Help Center. In the end, a metric with the same name is calculated differently depending on where you look. Google (official)
  • Account for known error periods when comparing long time spans. From May 13, 2025 to April 27, 2026, a logging error made impressions, and the click-through rate and average position that depend on them, inaccurate. Clicks were not affected. GSC Help Center

Double-check when you hear statements like these

  • “We’re at position 3 on average, but we get no inquiries.” You may be cited in an AI Overview, which only makes the number look good. The spot that actually gets clicks may be much lower, so check clicks and click-through rate side by side. Search Engine Journal
  • “The average went down, so our rankings dropped.” It may be the result of appearing for more queries. SEOTesting John Mueller said that if impressions are much lower than expected, you can assume the site appeared only briefly in part of the total impressions for that query. It is a signal to read the position number together with impressions, not on its own. Search Engine Journal
  • “This query slipped from position 5 to 9.” The average position of queries with very few impressions swings wildly because of localization and personalization, and it can differ greatly from what you see when you search yourself. It is safer not to treat changes in very low-impression queries as a signal. If you want a baseline, set it by looking at your own site’s impression distribution. SEOTesting
  • “Let’s make site-wide average position the headline metric in the monthly report.” This author’s advice is that the overall average leaves almost nothing to interpret. Practical Ecommerce Suggested alternatives are to split positions by result type and look at each one’s click-through rate, and to use percentile metrics such as the median or the 90th percentile instead of the average. Search Engine Journal

References

Aug.2026By Studio JT

More from Measurement

Like this term?
Give it a like!

(142)