Study
A higher fill rate did not mean more money
Two sites we run, trailing 30 days. The one that filled fewer of its ad requests earned more than double the page RPM.
"100% fill, any traffic" is the pitch from the networks we steer people away from. Fill rate is the number they sell, because it is the easiest one to move and it sounds like money. It is not. Here is the same metric on two sites we run, and the result is backwards from the pitch.
The two sites
Same owner, same method, same month. One is a broad consumer-interest site; the other serves a business audience. We have anonymized them, and page RPM is a per-thousand ratio, so none of this is a revenue figure.
| Site A | Site B | |
|---|---|---|
| Audience | Broad consumer | Business buyers |
| Ad requests filled | 83% | 41% |
| Viewable impressions | 44% | 64% |
| Page RPM | $1.54 | $4.11 2.7x |
AdSense, trailing 30 days. Site B filled half as many of its ad requests as Site A and earned more than double per thousand page views.
What fill rate actually measures
Fill rate is the share of your ad requests that came back with any ad at all. It says nothing about what those ads were worth. A request goes unfilled when no bid clears your price floor, so a lower fill rate often means the auction looked at the offers and declined the cheap ones. Site B filled 41% of its requests and the impressions it did serve were more viewable (64% against 44%) and far more valuable per thousand. The other 59% were not lost revenue. They were bids not worth showing.
Why the bad networks sell you fill
Getting to 100% fill is trivial if you stop being choosy. Accept pop-unders, pop-ups, floor-scraping CPMs, and every redirect, and every request fills. That is exactly the inventory behind the "100% fill, any traffic" networks we warn about. The fill rate goes up, the RPM and the reader go down, and the number on the dashboard looks healthier while the business gets worse. A high fill rate is not evidence of a healthy setup. It can be evidence of a floor set too low.
What to watch instead
Watch page RPM and viewability. Those are the two numbers that move with what you actually earn. Treat fill rate as a diagnostic, not a goal: a sudden drop can flag a real problem, like a broken unit or a misconfigured floor, and is worth investigating. But chasing the number upward, or letting a network loosen your floors to lift it, is optimizing the one metric that does not pay you. The lever is the same as everywhere else on this site. Is the ad seen, and is it worth showing.
How we measured it
AdSense ad request coverage, page RPM, and Active View viewability, trailing 30 days, on two sites in our network. Two sites is an illustration, not a proof, and a low fill rate can have benign or fixable causes. The point is narrow and it holds: a higher fill rate did not mean more money here, so fill rate alone cannot tell you how a site is doing. The numbers are real and our own.