Description: Why does Adsterra revenue stay flat when impressions rise? An anonymized publisher report reveals geographic differences, native ads, performance-based payouts and the gap between pageviews and counted impressions.
The answer: plenty of visitors, a strangely quiet cash register
A small multilingual site earned $0.11 from 1,208 ad impressions on one day, yet $0.80 from 39 on another. The latter day was still incomplete. Even so, around 31 times fewer impressions generated more than seven times the money. The spreadsheet appears to be telling jokes.
To understand it, separate the traffic source, the ad format, whether the ad was counted, and the action for which the advertiser paid.
1. Japan contributed 91.6% of impressions
Across five reporting days, the anonymized data showed 1,733 impressions and roughly $1.31. Japan contributed 1,587, or 91.6%, yet its reported effective yield was about $0.001 per thousand impressions. If that rate continued unchanged, even one million impressions would earn just $1 mathematically. That is not a benchmark for Japan's entire advertising market.
South Korea yielded approximately $0.13 from 41 impressions, Taiwan $0.13 from 26, and Mexico $0.75 from only two. Extrapolating Mexico to roughly $377 per thousand is misleading: one conversion or accounting adjustment can dominate two impressions.
2. Native Banner appeared to do all the work
A separate placement report showed Native Banner making $1.19 from 192 impressions. A mobile 320×50 banner brought in $0.05 from 942; a 300×250 unit brought in $0.01 from 417. Native produced about 92% of that report's $1.29 site revenue.
Should all banners be replaced? Not yet. If native units mainly reached overseas visitors while standard banners mostly reached Japanese readers, country and format changed together. Compare Japan/native, Japan/standard, overseas/native and overseas/standard separately.
Placement impressions add to 1,690, whereas country impressions total 1,733. Site-level revenue is $1.29 versus $1.31 in daily aggregates. Those differences in reporting scope or filters need resolution before declaring a winner.
3. Advertisers often pay for an action
Adsterra explained in January 2026 that many banner campaigns reward registrations, installs, purchases or other actions. Impression- and click-based campaigns also exist.[5] A thousand people seeing an advert can therefore generate almost no revenue without qualifying actions.
Think of a thousand passersby outside a store: they are not a thousand purchases. Inspect campaign payment models, clicks and conversions for each placement.
4. Pageviews are not the same as ad impressions
Another analytics report listed approximately 1,109 pageviews and 108 ad impressions. Dividing gives 0.097 ad impressions per pageview. Since time zones, time windows and ad-eligible pages may differ between systems, that's a diagnostic clue rather than proof of a broken installation.
Trace eligible pageview → ad request → loaded and counted impression → earnings. Ad-free pages, blockers, unviewed placements, slow scripts or a lack of available ads can all widen the gap.[6]
5. Would AdSense earn more?
Google defines page RPM as earnings divided by pageviews, times 1,000. Ad-impression RPM uses a different denominator.[1][2] Compare providers using equivalent geography, device, page type and period, focusing on revenue per pageview too.
At 1,587 impressions, a hypothetical $0.50-per-thousand yield gives roughly $0.79, and $1 gives $1.59. These are arithmetic scenarios, not AdSense forecasts. A spectacular percentage increase can still mean a small amount of cash.
Google permits other networks alongside AdSense subject to its rules.[3] That makes it possible to keep an effective native ad while testing a limited number of weak banners. Measure loading performance, reader experience and accidental-click risk as well.[4]
6. Check the register before widening the door
Align reporting periods, cross-tab country × format × device, measure losses from pageview to counted impression, then test another provider fairly. With small samples, a single conversion can flip the leaderboard.
The shop has customers, the lights are on, and the poster looks great. Before paying to widen the entrance, make sure the register is plugged in.
References (7)
[1] https://support.google.com/adsense/answer/112030?hl=en
[2] https://support.google.com/adsense/answer/112032?hl=en
[3] https://support.google.com/adsense/answer/9728?hl=en
[4] https://support.google.com/adsense/answer/1346295?hl=en
[5] https://adsterra.com/blog/how-banner-ads-make-money/
