Attributed vs incremental CPI and ROAS
Why the cheapest channel on the dashboard is not always the cheapest source of new users.
Incremental CPI is spend divided by the installs the ads caused, and iROAS is the revenue from those installs divided by spend. In the sample, Channel A's attributed CPI of $1.00 becomes $5.00 once only incremental installs count, while Channel B moves from $2.50 to about $3.57, so the ranking flips.
Channel A (brand search)
Channel B (social prospecting)
Sample data: a made-up app to show the method, not a benchmark.
| Sample month | Channel A | Channel B |
|---|---|---|
| Attributed CPI | $1.00 | $2.50 |
| Incremental installs | 4,000 | 5,600 |
| Incremental CPI | $5.00 | $3.57 |
| Attributed ROAS | 4.00x | 1.60x |
| Incremental ROAS | 0.80x | 1.12x |
The order flips. On attributed CPI, Channel A looks cheaper; on incremental CPI, Channel B is. Moving budget on attributed numbers alone would move it the wrong way. Channel A returns less incremental revenue than it costs (iROAS under 1x) in this sample.
Why the ranking can flip
Channels that reach people already close to installing (brand search, retargeting, some affiliate traffic) tend to show low attributed CPI and low incremental share. Prospecting channels that reach new people show higher attributed CPI but more of it is real. Dividing by incremental installs instead of attributed installs corrects for that.
The guide to acting on results shows how to turn these numbers into a budget move.
The maths
The incremental share for each channel comes from a test on that channel. Reusing one channel's result for another is a guess, not a measurement.
Questions
What is a good iROAS?
Above 1x means the campaign returns more incremental revenue than it costs over the revenue window you measured; below 1x it does not. The bar you need depends on margins and how long users keep paying, so set it from your own unit economics, not a benchmark.
Is incremental CPI always higher than attributed CPI?
Yes, or equal. Incremental installs are a subset of attributed installs, so dividing the same spend by a smaller number gives a higher cost per install.
Should I cut a channel with low incrementality?
Not automatically. First check the result's interval and whether the test covered a typical period. Then try reducing spend rather than stopping, and measure again: many channels have a level where the next dollar is still incremental.
What is incrementality in app marketing?
Incrementality is the share of conversions that would not have happened without the ad. It is measured by comparing people who could see the ads with a similar group who could not, so it captures what the ads changed rather than what they were credited with.
Is attributed the same as incremental?
No. Attribution credits a campaign when a user touched an ad before installing, even if that user would have installed anyway. Incremental installs are only the ones the ads caused, so they are always equal to or fewer than attributed installs.
How big should my holdout group be?
Big enough that the control group collects enough conversions to detect the lift you care about within the time you can wait. Lower conversion rates and smaller expected lifts need bigger holdouts or longer tests; the holdout simulator on this site shows the trade-off.
Other labs
- 01Holdout test simulatorPick a holdout size, an audience and a true lift, then watch a simulated test day by day until the confidence interval clears zero.
- 02Lift test readerEnter test and control results and get the lift, a 95% confidence interval, a plain-English verdict and the usual misreadings.
- 03Cannibalisation labSee how a paid push can be credited with installs that were going to arrive organically, and what that does to the organic line.