A decreasing RPM is seldom the real problem. It is an indicator arising weeks after the real issue occurred someplace else – whether in viewability, traffic quality, or a request channel discreetly disappearing. Publishers that only track the main revenue number end up treating the fever rather than determining the infection.
This is what is changing in the digital publishing industry today. Monetization previously involved one dashboard and one number: revenue, potentially broken down by day. Now it is a mechanism with many pieces, and the metrics that are important are those that tell you if your current visitors are effectively being monetized, and not only the number of visitors you have. Below are eight indicators that provide real answers, not only fulfill some reporting needs.
Session RPM, not page RPM
Many publishers still rely on Page RPM as a performance metric, but it can be misleading. When you increase the number of ad units on a page, the Page RPM might go up due to an immediate increase in revenue. However, if the page becomes too heavy with ads, it can lead to slower loading times, a decrease in user experience, and lower engagement, causing readers to leave the site after viewing fewer pages. In this case, the overall session value would decrease.
Session RPM measures revenue per 1,000 sessions instead of per 1,000 pageviews. It gives you a clearer picture of whether each visitor is more or less valuable to you. If your session RPM remains the same or decreases while your page RPM increases, it indicates that you are sacrificing long-term reader value for short-term profit by overloading your layout with ads. The solution here is not to add even more ad units but to optimize their placement and focus on reducing the number of ads so that your page speed and user experience are not compromised.
Viewability rate, tracked apart from served impressions
An ad impression that goes unseen is not as valuable as one that is viewed, and advertisers understand that. The MRC standard considers an ad viewable when 50% of its pixels are in view for one second for display and 75% of its pixels are in view for two seconds on larger units. A two-second minimum is used for video; that won’t change. If your viewability rates are significantly below those guidelines, your CPMs and fill rates are probably suffering, as demand-side platforms and buyers apply steep discounts to unseen or low-viewability impressions.
If your served CPM and your viewable CPM show a big gap, you don’t have a demand problem, you have a layout/ad placement/refresh slot issue. No amount of SSPs slicing and dicing your audience will overcome this. Auto-refreshing sidebar units on desktop can keep the ad server honest, but do little to build a relationship with the reader. Auto-refresh must be used judiciously on premium sites, counting only “human” impressions – repeating ads to the same user does not make a site premium. Viewability is an initial building block to determine if you can actually use the programmatic marketplace to sell your audience or if direct sales, while they’re still out there to be had, are your only chance to monetize. If viewability is 60% or lower for a reasonably long campaign, a buyer might just walk away at billing and never come back.
Fill rate read against yield, never alone
On its own, fill rate doesn’t really mean much. Focusing on optimizing a fill rate cannot guarantee higher revenues. It merely indicates how successful a publisher is at selling their ad inventory. A fill rate is simply a ratio of the number of ads requested divided by the number of ads returned by the ad server in response to those requests. A fill rate of 100% means all ad requests are filled – but this does not mean that all inventory is sold. A 100% fill rate at a low price isn’t really a good thing; it just means you are giving away all your inventory at a low price! Marketers doing indirect sales or looking to add header bidding to your mix, focus on optimizing for revenue, not fill rate.
This is where unbought impression share becomes important. Segmenting by device usually surfaces problems fastest: mobile web fill often lags desktop and app because of smaller ad formats and weaker identity signals. This is also where the choice of demand partner starts to matter more than most publishers admit. A well-chosen ad network for publishers will give you session-level visibility into fill and price by channel, rather than a single blended number that hides where the actual leakage is happening.
eCPM segmented, not blended
Blended eCPM is what publishers tell management, and the number that tells them the least. It averages together your best-performing inventory and your worst, and the average always looks more stable than the reality underneath it.
Segment eCPM by direct-sold, PMP, and open auction, then again by geography and device, and a pattern shows up almost every time: a small share of inventory – often direct and PMP deals in top-tier geos – produces a disproportionate share of total revenue. The rest of the auction, particularly long-tail geo and older device traffic, is often clearing at a fraction of the price. This isn’t necessarily bad news. It tells you where to focus sales effort, where to raise floor prices without losing fill, and where open auction demand is simply too thin to be worth much attention. If you’ve never broken out eCPM this way, do it before your next rate-card conversation with a demand partner – it’s the single fastest way to find out if a “revenue decline” is actually a decline in one narrow segment dragging the average down.
Engagement depth as a monetization number
Metrics such as pages per session, scroll depth, and engaged time on page are typically found on the editorial dashboard. However, they are just as important for the revenue dashboard because it is these metrics that determine what proportion of ads a reader is exposed to. Two pages read at halfway down the page give less exposure to ads and overall creates less ad revenue than viewing four pages and the entire article. These figures thus determine the theoretical maximum of how much revenue we can get for a visit. This includes whichever categories of ad models we wish to use. Ads bought based on the amount of time an ad is in view will be self-limiting if scroll depth is small. Similarly, high viewability won’t help ad revenue if all users read one page and leave.
First-party data and consent coverage as a revenue multiplier
Every time access to a third-party signal or data source is lost, and in the absence of first-party identifiers, a login, or consent, any money that leaks away from publishers goes to the buy side. That means that for all the talk this year of rebuilt privacy-compliant identities at the scale programmatic buyers are used to, the majority of the bleeding-edge programmatic and any added automation of DOOH, CTV, or audio buying will have a first-order effect of lowering floor CPMs for the vendors involved.
Report your first-party data coverage rate and consent rate as revenue metrics, not compliance metrics. If only 40% of your sessions carry a usable identifier or valid consent signal, you’re likely leaving a meaningful eCPM premium on the table across the other 60%. Publishers investing in login walls, newsletter capture, or identity solutions like UID2 or RampID are doing it as much for monetization as for audience relationship – the addressability layer is now a direct input to price. If this number isn’t on your dashboard next to RPM, you’re missing the metric most likely to explain a slow, steady decline in average price across identical traffic volumes.
Invalid traffic rate by acquisition source
All pageviews don’t have the same value, and some have none at all. Invalid traffic (IVT) and made-for-advertising (MFA) inventory might increase your pageview estimates, but they’re simultaneously reducing your average eCPM, and long-term, undermining buyer confidence in your domain. A highly effective traffic source by the simplistic pageview number in your dashboard might, in fact, be the exact source dragging your blended price down.
The diagnostic here is straightforward: break IVT rate out by acquisition channel – direct, search, social, referral, paid – rather than looking at a single site-wide number. A spike in one channel is usually easier to isolate and cut off than a site-wide IVT problem, and it answers a very practical question: is this traffic source worth paying to acquire? Ads.txt, sellers.json, and the SupplyChain Object exist precisely so buyers can trace where an impression actually came from and verify it against a declared seller list. A transparent, well-run ad network will surface IVT by source rather than leaving publishers to infer it from a falling blended eCPM after the damage is done.
Watching the right fever chart
None of these eight metrics replace RPM as the number that ends up in a board deck. What they do is explain it. A publisher tracking session RPM, viewability, segmented fill and eCPM, engagement depth, identity coverage, invalid traffic, Core Web Vitals, and revenue concentration has a working diagnosis for almost any revenue movement before it becomes an emergency. The publishers still watching one number are the ones who find out something broke only after it’s already cost them a quarter.
