What Average Position in Search Console Actually Measures

When you review organic search performance in Google Search Console, average position summarizes where Google showed your result across the impressions included in the report. An impression is a recorded appearance of a result in Google Search. The position is an average of those recorded appearances, so it does not tell you the exact rank a person will see on a particular search.

Google records the topmost position occupied by your property or page for each impression. Use the average to track trends. Its meaning depends on the report settings and on how Google displayed the results. Those details help you investigate changes without treating the average as a fixed rank.

What Search Console averages

Google assigns positions to items on the search results page. These items include standard links and features such as carousels or knowledge panels. A feature, or other result item, can contain several links, and those links can share one position. Google counts these items in the page layout. As a result, the same position number does not always mean the same visual location or the same number of links on screen. Google also says ads do not occupy a search position. Its explanation of how position is recorded describes these rules and notes that the counting method can change.

Search Console averages the recorded positions across impressions. That makes the average impression-weighted: results with more impressions contribute more to the number. For example, if one impression is recorded at position 1 and 99 are recorded at position 10, the average is 9.91. This is an illustration, not observed Search Console data.

Google records a position only when the result receives an impression. If a result appears on a later results page that the searcher does not view, Google says it does not record a position for that result for that search. The metric therefore describes observed appearances, not every query for which a page could potentially rank.

Property and page averages use different units

How Search Console groups results changes what “topmost result” means. For a property, it uses your site’s highest-positioned result for a query. For a page, it reports each URL separately.

If three pages from your site appear at positions 1, 2, and 3 for a query, the property-level view records only the topmost result at position 1, while the page-level view records each URL at its respective position. Google’s reporting documentation shows why these averages can differ.

Check whether a report is grouped by property or page before comparing averages. A site-wide property average and a page’s average do not describe the same set of results.

Why an average changes

A lower numerical average generally means a higher position within the report’s selected scope. But a change in the average does not prove that a particular page moved up or down for a particular query.

The average can change when positions change for impressions already in the report. It can also change when the report includes a different mix of impressions. For example, the report may include more impressions from queries where your site appears lower, even if its position for a high-priority query stayed the same. A different mix of pages or devices can also change the aggregate. The same applies when the mix shifts across countries or search result formats.

Start by checking the report’s property, search type, dates, and filters. Then compare equivalent periods and look at clicks and impressions alongside position. Filter to the page or query you need to diagnose, and compare device or country segments when those differences matter. Google’s common Performance report tasks cover these kinds of comparisons.

A manual Google search can help you inspect a result layout, but it cannot reconcile a report average. Your search can differ in location or search history, and Search Console averages many recorded appearances. Treat the manual search as a spot check.

Common reporting failures and limits

A visible query table does not necessarily contain every query behind the chart totals. Google omits some queries for privacy and may leave out rows because of internal data limits, so the table may not explain the full chart total even when the report is working as intended. Google describes these query and row limitations.

Chart and table figures can also differ because of aggregation, privacy filtering, or row limits. Recent data may not be complete; Google says Search Console data is usually available after a delay. Before treating a small discrepancy or a recent dip as a ranking change, check the selected date range and whether the latest data is complete. Google lists these issues in its troubleshooting guidance.

Average position does not explain why a result changed or establish whether the change caused a gain or loss in clicks or made a page more or less useful to searchers. Google recommends assessing performance through trends in clicks and impressions as well as position. If those metrics change, investigate the affected pages and queries rather than treating the average as a diagnosis. Google’s guidance on debugging traffic drops provides a starting point.

How AI Overviews affect position

Google says an AI Overview occupies one position in Search, and all links within that overview receive that same position. The number therefore identifies the overview’s position in the results, not a separate rank for each cited page. An average that includes AI Overview impressions cannot tell you which link appeared more prominently within the overview.

Use the same caution as with other result formats: check the reported position alongside clicks and impressions, and avoid interpreting it as a complete measure of visibility or traffic impact.

Key Takeaways

• Read average position as an aggregate: It averages recorded positions across impressions in the report’s selected scope.
• Check the aggregation level: Property averages use the site’s topmost result; page averages treat URLs separately.
• Investigate changes by segment: Compare equivalent periods, then filter by page, query, device, or country and review clicks and impressions.
• Treat the number as a diagnostic signal: It does not reveal an exact user-specific rank, visual prominence, or the cause of a change.
• Account for incomplete reporting: Privacy filtering, row limits, and data delays can make query tables or recent totals incomplete.

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com