In programmatic ad mediation, eCPM (Effective Cost Per Mille) is the universal unit of measurement used to evaluate the monetization efficiency of ad inventory. While "mille" is Latin for one thousand, eCPM does not represent a static price list. Instead, it is a retrospective calculation of the net publisher revenue generated per 1,000 rendered ad impressions across mixed auction models (CPC, CPI, CPM, and CPA).
Developers who optimize strictly for a higher dashboard eCPM without accounting for fill rate variance, request-to-impression drop-off, and user session frequency routinely reduce their actual bankable payouts. This reference breaks down the exact formulas, metric distinctions, and floor-pricing trade-offs required to maximize total net yield.
1. Core eCPM & Reverse Revenue Formulas
Ad networks calculate eCPM by taking your net finalized earnings (after the demand partner's revenue share has been deducted) and dividing by the count of billable impressions where the creative was successfully rendered on the client device.
Worked Calculation: Suppose a mobile puzzle game serves 420,000 rewarded video impressions and 1,150,000 banner impressions over a 30-day billing period. The rewarded video slot earns $6,720.00 and the banner slot earns $690.00.
- Rewarded Video eCPM:
($6,720 / 420,000) × 1,000 = $16.00 - Adaptive Banner eCPM:
($690 / 1,150,000) × 1,000 = $0.60 - Blended App eCPM:
($7,410 / 1,570,000) × 1,000 = $4.72
Notice how high-volume banner impressions pull the blended eCPM down from $16.00 to $4.72 even though banners added $690.00 in pure incremental profit. Never judge an application's health by its blended eCPM alone; always segment eCPM by placement format and geography.
2. Why Non-CPM Campaigns Still Report an eCPM
Over 75% of mobile gaming and utility ad buyers do not bid on a pure brand awareness CPM basis. Instead, User Acquisition (UA) managers bid on CPI (Cost Per Install), CPA (Cost Per Action / Day-7 ROAS), or CPC (Cost Per Click). Mediation engines translate these performance bids into an equivalent eCPM in real time using historical conversion probability:
For example, if an advertiser bids $3.50 per install (CPI) on AppLovin or Unity Ads, and machine learning models predict that your placement has a 3.0% click-through rate (CTR) and a 25% click-to-install conversion rate (CVR), the bidder enters the auction at: $3.50 × 0.03 × 0.25 × 1,000 = $26.25 gross eCPM. If your app suffers from accidental clicks or low-intent users who never install advertised apps, the bidder's predicted CVR drops, immediately collapsing your eCPM.
3. Metric Distinction Matrix: CPM vs eCPM vs RPM vs ARPDAU
Reporting dashboards across Google AdMob, Google AdSense, AppLovin MAX, and Unity LevelPlay use overlapping terminology. Engineers and financial analysts separate these metrics by their denominator:
| Metric | Exact Formula | Side of Market | Primary Operational Use |
|---|---|---|---|
| CPM | (Ad Spend / Impressions) × 1,000 | Buy-Side (DSP) | Contractual cost paid by an advertiser before exchange fees. |
| eCPM | (Net Revenue / Filled Imps) × 1,000 | Sell-Side (App/Web) | Comparing yield per individual ad unit or mediation bidder. |
| Page RPM | (Net Revenue / Pageviews) × 1,000 | Web Publishers | Total earnings per 1,000 page loads across all on-page ad slots. |
| rCPM | (Net Revenue / Ad Requests) × 1,000 | Mediation Architects | Penalizes unfilled requests; exposes false high-floor eCPMs. |
| ARPDAU | Daily Net Ad Revenue / DAU | Product & UA Teams | Benchmarks user daily value against user acquisition CPI cost. |
4. The Price Floor vs. Fill Rate Optimization Curve
In both waterfall configurations and OpenRTB header bidding, publishers can configure a minimum clearing price (an eCPM floor). If no buyer bids at or above the floor, the ad request returns a NO_FILL response.
Setting an overly aggressive floor price is the single most common cause of self-inflicted revenue loss. Consider the empirical data below from a 500,000 daily request interstitial placement in Western Europe:
| Configured Floor | Ad Requests | Fill Rate (%) | Delivered Imps | Reported eCPM | Net Daily Revenue |
|---|---|---|---|---|---|
| No Floor ($0.01) | 500,000 | 94.0% | 470,000 | $4.80 | $2,256.00 |
| Optimized Floor ($3.50) | 500,000 | 81.0% | 405,000 | $6.30 | $2,551.50 (Peak) |
| Aggressive Floor ($8.00) | 500,000 | 38.0% | 190,000 | $9.40 | $1,786.00 |
| Extreme Floor ($15.00) | 500,000 | 11.0% | 55,000 | $16.80 | $924.00 |
At the $15.00 floor, the publisher boasts a $16.80 eCPM—yet earns $1,627.50 less per day than the $3.50 floor configuration because 89% of ad requests went unfilled. Moderate floors ($3.50 above) work because they force second-price or shaded first-price DSP shading algorithms to bid closer to their true valuation without starving the inventory of fill.
5. Impression Pacing & Session Decay (Why First Impression Pays Most)
eCPM is not constant across a user's daily session. In programmatic auctions, the first impression shown to a user each day (Impression Depth #1) clears at the highest eCPM because all demand partners are competing for their first daily exposure to that device ID.
- Impression #1 of Session: Typically clears at 1.45x to 1.80x the daily average eCPM.
- Impressions #2 to #4: Clear near the baseline average eCPM (1.00x).
- Impressions #8+: Clear at 0.40x to 0.65x the baseline due to DSP frequency caps (advertisers cap bids after showing the same creative 3 times to one user in 24 hours).
Model Your Own Blended eCPM & Fill Rate
Test how fill-rate changes and geographic tiers impact your monthly payout in our interactive calculator.