Average position dropped.
Impressions increased.
Your Search Console dashboard shows a worse average position alongside more impressions and more clicks. That can be normal, misleading or a real problem. Here is how to tell the difference, using our own property data as the example.
More queries.
Lower average.
In equal 28-day windows (27 September to 24 October 2026 vs 30 August to 26 September 2026), Tendou's own Search Console property (Web search, worldwide, final data read on 26 October) showed:
| Metric | Earlier 28 days | Later 28 days |
|---|---|---|
| Clicks | 7 | 44 |
| Impressions | 49 | 1,465 |
| Average position | 25.1 | 63.7 |
At first glance that looks like a severe ranking decline: position moved from 25 to 64. But the earlier period had only 49 impressions across the entire property, which is too small to establish a stable historical baseline.
In Singapore specifically, impressions rose from 26 to 908 while clicks rose from 2 to 11, and average position moved from 37.1 to 72.6. These Singapore-only totals show more exposure at poor average positions. They are a separate geography from the worldwide table above, not extra impressions to add to it.
Averaging explains
the headline number.
At property level, Search Console records the position of the highest link to the property for each search impression, then averages those recorded positions. New impressions at positions 50 to 100 can worsen that average. That does not establish whether an existing query also lost position; compare those queries separately.
Lower position numbers are better. Property and page grouping count impressions differently, so a weighted average of query-page rows is not a substitute for the property-level figure. Google explains position and aggregation.
The commercial weakness is real. For the exact query aeo agency singapore, the worldwide average moved from 28.8 across 6 impressions to 69.4 across 183 impressions, with no clicks in the later period. The small earlier sample limits the comparison, but the later poor position still needs attention. Broader query coverage and worse averages on existing queries can happen together; these figures do not establish what caused either change.
Separate the
old from the new.
Before concluding that rankings declined, check whether the position change comes from existing queries worsening or from new queries appearing at low positions:
- Filter by individual query: compare the same query across both periods. If specific queries you care about lost position, that is a genuine ranking change worth investigating.
- Check query count: new low-ranking queries can worsen the aggregate. This can coexist with declines on existing queries. Query exports also omit anonymised searches, so they do not reconstruct all property impressions.
- Watch for small denominators: an average position calculated from 6 impressions (as our AEO Singapore query had in the earlier period) is not a reliable baseline. A move from 28.8 to 69.4 sounds dramatic, but the earlier number was based on 6 data points.
- Compare property totals with page-filtered totals: these use different aggregation methods in Search Console. Property aggregation generally counts the site once for a recorded search impression, while page aggregation can count separate pages. Use the same grouping in both periods; totals can differ.
Clicks are not
enquiries.
More impressions and more clicks are positive signals, but they are not confirmed enquiries or revenue. A click means someone saw a search result and tapped it. Whether that visitor contacted the business, purchased something or left immediately requires separate measurement.
Similarly, average position improving or worsening does not by itself explain why AI answers do or do not name a business. Google says existing SEO best practices remain relevant to its AI features, but search position and AI recommendation are measured separately. Google's AI features guidance.
Wait for
final numbers.
At the 26 October read, Search Console marked 25 October as incomplete. We excluded it from both final-data windows. Preliminary figures can change as processing finishes; do not compare a partial day with a complete one.
When reading any Search Console comparison, confirm that both periods use final data. Google's documentation notes: "Data for recent dates may be preliminary." Check the data freshness indicator in your Search Console account.
Read the average
in context.
A worse average position alongside more impressions can reflect new exposure, declines on existing queries, or both. The correct response is to check individual queries, verify the earlier baseline was large enough to be meaningful, and measure business outcomes separately from search metrics.
In our case: property impressions grew from 49 to 1,465, clicks from 7 to 44, while average position moved from 25.1 to 63.7. The earlier 49-impression baseline was too small to establish a reliable ranking level. The same AEO Singapore query also had a worse average in the later window. The aggregate alone cannot separate exposure changes from losses on existing queries, establish a cause, or prove that a release worked.
Get found on Google.
Get recommended by AI.
Send us your website and the services you want found. We check Google and four AI engines on the same questions before proposing any work.