Let more buyers compete
Every extra bidder is another chance at a higher price. Through VisualAds, Google’s demand, Open Bidding and header bidding partners bid in the same auction.
Ad revenue optimization
Ad revenue optimization: what actually moves the number
Ad revenue comes down to how many impressions you sell and what each one earns. These are the levers that move both — and the ones VisualAds takes care of for you.
- Demand partners
- 12, in one auction
- Reporting
- Real time
- Revenue share
- 5% · you keep 95%
- Payout terms
- Net 30, monthly
The basics
Four numbers worth watching
Revenue = impressions × eCPM ÷ 1,000. Everything below works on one of these.
-
Volume
Impressions
How many ads were actually shown. Grows with traffic and with ad slots that users really see.
-
Demand
Fill rate
The share of ad requests that returned a paid ad. Low fill usually means too few buyers, or requests buyers do not want.
-
Price
eCPM
Revenue per 1,000 impressions. It reflects how buyers value your audience, format and placement.
-
Result
Revenue
The result. Improve one number without quietly lowering another.
The levers
Six levers publishers control
No tricks — just the things that consistently make a difference.
Place ads where people look
Ads that never appear on screen rarely earn. Put units where users actually scroll and read, and do not stack several at the top of the page.
Use formats that fit
Display, video, rich media, VAST in-stream and native suit different placements. Video and native often earn more where they fit the content.
Keep ads.txt complete
Many buyers skip inventory they cannot verify. Publish every ads.txt or app-ads.txt line your dashboard shows and keep the file reachable.
Stay policy-clean
Policy issues can limit or stop ad serving. Fix anything reported in your Policy center quickly, and never buy or reward clicks.
Measure, then change one thing
Compare eCPM and revenue by ad unit and property in your reports. Change one variable at a time and give it at least a week.
Who does what
What VisualAds handles, and what stays with you
Optimization is shared work. This is where the line sits.
-
VisualAds
Demand and auction
Your inventory is connected to Google Ad Manager demand, Open Bidding and header bidding partners.
-
VisualAds
Invoices and payouts
Invoices are generated automatically; payouts are monthly on Net 30.
-
Shared
ads.txt / app-ads.txt
Your dashboard shows the lines and their status; you publish them on your domain.
-
You
Placement and layout
Where ad units appear in your pages and screens, and how many there are.
-
You
Content and traffic
Following the Google Publisher Policies; issues are listed in your Policy center.
Avoid these
Mistakes that cost revenue
Most avoidable revenue losses come from a short list of causes:
- More ad units than the page can carry — extra slots add impressions but often pull average eCPM down and hurt the user experience.
- Ads next to buttons or navigation — accidental clicks are filtered as invalid and can lead to policy action.
- Paid or incentivized traffic that brings visits but no real attention.
- Judging a change after one day — eCPM moves with the day of the week and the season.
- A stale ads.txt or app-ads.txt after moving domains or changing the developer website in your store listing.
New to programmatic? Start with how the real-time auction works.
FAQ
Questions publishers ask
There is no single benchmark: eCPM depends on audience country, device, format, placement and season. Compare your own eCPM over time and between ad units rather than against averages you read online.
Only up to a point. Extra units add impressions, but units users rarely see lower the average eCPM and can hurt the user experience. Add them one at a time and measure.
Many buyers only bid on inventory they can verify through ads.txt (websites) or app-ads.txt (apps). Missing or outdated lines mean fewer bids.
By bringing Google’s demand, Open Bidding and header bidding partners into one auction, and by giving you real-time reports on impressions, clicks, eCPM and revenue.
Give each change at least a full week of data, so weekday and weekend patterns are both included, before you compare it with the previous period.