Payments are experiencing issues due to temporary restrictions in Russia. If your payment does not go through, please submit a support request.Our support team is available 24/7 — we are always here to help with hosting and server issues.We are now accepting requests for dedicated server rental and colocation services in our data center.Reminder: we recommend enabling backups for additional data protection.A new VPS/VDS lineup with NVMe storage and improved performance is now available.Maintenance work on some servers has been completed. All services are operating normally.
Article5 min readViews6

Average Position in Search Console Changed: How to Understand What Happened

I analyze Google's average position by individual impressions and demonstrate via a sample calculation how device composition can shift the overall metric without movement within segments.

Comments 0

A magnifying glass and scales on the desk as a metaphor for checking an averaged metric.
In this article

In the sample report, the site's average position yesterday was 3.4, and today it became 6.7. The immediate reaction is understandable: the site has dropped. However, such a conclusion is premature. First, you must verify which impressions were used to calculate each average: queries, pages, devices, countries, and search type may have changed, even if the position remained the same within a comparable segment.

Quick guideline: read the position number in Google Search Console together with impressions and clicks, and compare only after applying identical filters and for comparable periods. A single overall metric helps detect a change but does not explain its cause.

What Exactly Does Search Console Average?

In the Google Search performance report documentation, average position is defined as the average rank of the site's top result. On the chart, data is aggregated across the entire Search Console property, while in the table, it is broken down by the selected row: query, page, country, device, or date.

If a site appears at positions 2, 5, and 8 for a single query, the top position—2—is used for that impression. If the top result for another query is at position 3, the average for the two impressions is 2.5: (2 + 3) / 2. The site's second and third links do not turn one impression into three position observations.

Training example: with one impression of the first query, position 2 is counted; with one impression of the second, position 3 is counted; their average is 2.5.
Sample example based on Google's methodology: for each impression, the site's top result position is taken.

There is a nuance: a position is recorded only for an impression. If the result did not appear on the page the user opened, or if the visibility condition for its format was not met, the impression is not logged and no average position value exists. This follows the methodology for counting impressions, clicks, and position. Therefore, a dash indicates that no position was recorded in the selected segment, not automatically that the position is 101, that there is no demand, or that the page was removed from the index.

How impression composition affects the average

Let's take a training example. In the first period, the page received four impressions on desktops at position 2 and one impression on a smartphone at position 9. Average position: (4 × 2 + 1 × 9) / 5 = 3.4.

In the second period, the position within each device remained unchanged: 2 on desktops and 9 on smartphones. However, there were now two impressions on desktops and four on smartphones. The new average: (2 × 2 + 4 × 9) / 6 = 6.7 after rounding. The overall metric worsened, even though neither of the two segments shifted.

This does not prove that devices explain the site's actual data. The example illustrates the mechanism: the average depends on both position and the weight of each segment. The same effect is possible when the set of queries, countries, pages, or types of search results changes.

Training comparison of two periods: positions within devices remain unchanged, yet the average worsens from 3.4 to 6.7 due to the share of mobile impressions.
Training data, not site measurements: positions within devices are identical in both periods; only the composition of impressions changes.

Verification without guessing the cause

  1. Select the same search type and two comparable periods. For weekly businesses, comparing Monday with Sunday often introduces seasonality instead of providing an answer.

  2. Keep impressions, clicks, and average position on the chart. CTR is useful as the ratio of clicks to impressions, but by itself it does not explain why demand or the composition of search results changed.

  3. Start with a query or page showing a noticeable change, then sequentially break down the data by page, device, country, and search appearance. This is a separate filter, not a search type like Web, Image, or Video. Change one dimension at a time.

  4. Compare the position within identical segments and their share of impressions. If values within segments remain stable but shares change, the overall metric may have shifted due to composition.

  5. If a position, impressions, and clicks have deteriorated within the same segment and the change persists, proceed to check specific pages, queries, and technical events. A temporal match does not prove causation.

Google separately recommends applying a page filter to the position history, selecting a comparable period, and paying more attention to trends in impressions and clicks than to position alone. The step-by-step procedure is available in the official report scenarios.

When the problem is indeed with the page

Data composition is only one of the checks. If a specific query and page on the same device in the same country show a consistent decline in position along with impressions or clicks, examine the actual search results and changes to the page itself. It is useful to check whether the landing page has changed, whether another site URL has appeared, or whether the search intent has shifted.

If the report collects data on a different canonical URL, this is a separate issue. First, match the exact page pair and signals, as in the analysis of Google choosing a different canonical page. Moving data to a canonical address does not equal an automatic drop in the position of the original URL.

What cannot be concluded from this data

  • Search Console data pertains to Google. It cannot be treated as Yandex search data or equated with positions from another verification service.

  • Position is the relative rank of a result, impressions are recorded link appearances, and clicks are transitions from search. None of these metrics alone equals visits, orders, or revenue.

  • The latest data may be preliminary. Comparing an incomplete day with a completed period yields a quick but unreliable conclusion.

  • Some search queries are hidden for privacy reasons, and tables have row limits. The sum of visible search queries may not match the chart total.

  • Manual verification of an individual search query depends on time, location, device, and user history and does not have to match the report average.

Therefore, the solution does not begin with the phrase 'position dropped,' but with a precise slice: which query, which page, which device, country, search type, and time period. If both the position and the useful search result have worsened within such a slice, there is a subject for diagnosis. If only the overall average has changed, first recalculate the composition of impressions—sometimes it honestly answers a different question than the one asked.

Discussion 0

Share your experience and ask questions. Comments without links appear after editorial review.

No comments yet. Start the discussion.