Ad density
We added more ads. Here's what we still don't know.
A sweep raised our ad density by half on a site we run. The honest answer to "did it earn more" is that at this traffic, we cannot yet tell. That uncertainty is the finding.
By Mario Bailey · Published
Most advice about ad density is stated with a confidence the data does not support. Add a unit and revenue goes up, or add a unit and readers flee. On a small site, over a few weeks, neither claim is measurable, and the honest move is to say so. Here is a real density change on a site we run, and why we are not yet claiming it worked.
What changed
A July ad sweep raised the site's ad impressions per page view from 1.15 to 1.73, a 51 percent increase in density, over the same weeks that page views fell 7 percent. So there was more ad load on slightly less traffic.
| Before | After | |
|---|---|---|
| Ad impressions per page view | 1.15 | 1.73 +51% |
| Page views | baseline | -7% |
| Did revenue rise? | Unknowable at this volume | |
Density is a clean, measurable structural change. Its revenue effect at a few hundred page views a day is buried under click luck.
Why we cannot answer the obvious question
The structural change is real: more ad impressions per page view is a fact, not a coin flip. The revenue answer is the coin flip. At a few hundred page views a day, daily and weekly RPM swings several dollars on a single click, which is a bigger effect than the density change could plausibly produce over these weeks. Any "revenue is up" or "revenue is down" read right now is measuring which days happened to catch a good click, not the density. We would need a longer window, at least a full trailing 28 days, before the density's effect rises above that noise.
The thing we are actually watching for
The failure mode of adding density is not immediate. It is rising density with falling page RPM: more ads each earning less, because past a point viewability drops, the reader tunes out, and each extra unit competes with the ones already there. That pattern takes weeks to separate from noise, so that is the pair we track, density against page RPM over a trailing window, rather than celebrating the density number on its own. More inventory is not more revenue once viewability falls, and the only way to know which side of that line you are on is to wait for the window to fill.
Why we publish the uncertainty
Because the alternative is the entire problem with monetization advice. It is easy to add ads, watch a good week, and declare the change a win, then carry it into a site where it quietly costs viewability and rankings. Holding density near one ad per page view is our default precisely because the upside of going past it is unproven at the volumes most publishers run, and the downside is a real experience cost. When the window fills we will update this page with the verdict, whichever way it lands.
How we measured it
AdSense ad impressions per page view and page views before and after a placement change on a site we run. Density is a ratio and page views are indexed to a baseline, so nothing here is a revenue figure. The finding is deliberately incomplete: the structural change is measured and real, the revenue outcome is not yet separable from click noise at this volume, and claiming otherwise would be exactly the mistake this site exists to avoid.