eCPM in Plain English, With Worked Examples

eCPM is the average amount you earn for every 1,000 ad impressions your app or site shows. It is the number most ad dashboards put near the top, and the one most often misread.

The trouble is that your income is not paid in impressions. Some ads pay when they are seen, some when they are clicked, some when someone installs a game. eCPM squeezes all of that into one comparable figure, which makes it handy for planning and easy to misuse.

This guide shows the formula, a worked example, how eCPM differs from the other acronyms in your reports, and what moves it. Figures use the same benchmark tables as the revenue calculators, and every dollar amount is an estimate.

The formula, with one small example

eCPM = (total earnings / impressions) x 1,000

Say your app earned $37.80 yesterday from 21,000 impressions. Divide: 37.80 / 21,000 = 0.0018 dollars per impression. Multiply by 1,000 and the eCPM is $1.80.

You can run it in reverse to forecast. If you expect 50,000 impressions next month at a $1.80 eCPM, the revenue is 50,000 / 1,000 x 1.80 = $90. Every planning tool, including the AdMob revenue calculator, is built on that one relationship: impressions times eCPM, divided by 1,000.

Why the word is "effective"

Advertisers buy in different ways. Some pay a fixed price per thousand views (CPM). Some pay per click, and some pay only when a user installs something. As a publisher you only care what each batch of 1,000 impressions turned into, whatever the advertiser's pricing model was.

The "e" stands for effective. It converts every pricing model into the same unit after the fact. A click-based ad that earns $2.00 across 1,000 impressions has an effective CPM of $2.00, exactly like a fixed-price ad that was sold at $2.00. Because it is calculated from results rather than quoted in advance, it can change every day.

eCPM, CPM, RPM, CPC and CPI side by side

These five get mixed up constantly. The table separates them.

CPM, CPC and CPI are what advertisers pay. eCPM and RPM are what you actually got. RPM differs from eCPM by its denominator: pageviews or sessions rather than ad impressions. A page showing four ads makes four impressions, so the RPM of a page is roughly the sum of what its slots earn per thousand loads. If each of four slots earns $2.50 per 1,000 impressions, page RPM is about $10. See website RPM by niche for how that plays out for sites.

Converting is simple. A $0.40 CPC with a 0.5% click rate gives 1,000 x 0.005 = 5 clicks, or $2.00 per 1,000 impressions. A $2.00 CPI with 2.5 installs per 1,000 impressions gives $5.00, before the network takes its share.

Four things that move your eCPM

Format matters first. In our AdMob benchmarks for Tier 1 audiences, rewarded video is $12.00, interstitial $8.00 and banner $0.90. A banner impression is cheap to serve and worth a tenth of a rewarded one.

Geography is next. The same rewarded video benchmark falls from $12.00 in Tier 1 countries to $5.00 in Tier 2 and $1.40 in Tier 3. Your player's location often explains more than your ad setup does.

Platform and category follow. iOS applies a 1.3x multiplier in our tables, hypercasual games 0.8x, mid-core and strategy games 1.35x, and finance apps 1.6x. On the web, niche plays the same role. The mobile eCPM benchmarks break these out in full.

Season is the fourth. Advertiser budgets usually swell in the last quarter of the year and thin out in January, so an October-to-December rise followed by a dip is normal and not a sign that something broke. Treat any benchmark as a band, not a point: our range runs from 0.7x to 1.35x around the central figure.

A full example with three formats

Take a Tier 1 heavy game using the AdMob figures above on one day:

• Rewarded: 2,000 impressions at $12.00 earn $24.
• Interstitial: 3,000 impressions at $8.00 earn $24.
• Banner: 20,000 impressions at $0.90 earn $18.

Total earnings are $66 from 25,000 impressions, so the blended eCPM is 66 / 25,000 x 1,000 = $2.64. Now add 10,000 more banner impressions, worth $9. Earnings climb to $75, but impressions reach 35,000, and the blended eCPM drops to 75 / 35,000 x 1,000, about $2.14.

Revenue rose while eCPM fell. That is the single most important thing to understand about the metric: it is an average, and averages shift when the mix changes. Add cheap impressions and the figure sinks; remove them and it climbs, even if total income does the opposite.

Mistakes that make eCPM misleading

The commonest one is comparing a blended eCPM between two apps, or two weeks, with different format mixes. Compare format by format, and country by country, instead.

A second mistake is ignoring fill. A network might show a handsome eCPM while filling only half your requests, so you leave money on the table. Look at eCPM, fill and impression count together. Third is chasing a single high-eCPM day. Small samples are noisy, so judge in weeks.

Others: treating estimated earnings as final (networks adjust them before payment), confusing RPM with eCPM when moving between app and web reports, and cutting ad frequency purely to raise eCPM. Fewer impressions at a higher average can still earn less.

How to use the number

Use eCPM as a diagnostic, not a goal. When revenue moves, split it into impressions and eCPM first. If impressions fell, look at users, fill or placements. If eCPM fell, look at geography, format mix or season.

Then use it to plan. Multiply expected impressions by a realistic eCPM, using the low and high ends of the band, and get a range. Mediation can lift the figure by adding competing buyers, and rewarded or interstitial decisions shape the mix. For sites, try the AdSense revenue calculator. The AdMob help centre and AdSense help centre show how each reports its own figure, and our methodology explains the estimates.

A quick glossary for the rest of your reports

If a column in your dashboard still looks odd, the A to Z ad glossary covers fill rate, ARPDAU, session RPM and the other terms that sit next to eCPM. Read it alongside your first full month of data, once the numbers have settled and the estimates have been finalised.