What is iROAS (incremental ROAS)?
How is iROAS calculated?
Take the revenue that would not have happened without the ads. Divide it by what you spent.
Same spend. Same ads. The difference sits in the numerator.
Where does the incremental number come from?
From measurement, not from a dashboard. A holdout test is one way: some people or regions do not see the ads, and the gap in results is the effect. We use the term "iROAS" only for a figure that comes from a test or a calibrated model. A guess dressed up with an "i" does not count.
Sources argue about what belongs in the denominator1. So state what spend you used.
Is iROAS the same as platform ROAS times a factor?
Not by definition. It is a conversion. Common Thread Collective's ad spend benchmark applies it as iROAS = platform ROAS × incrementality factor2. The factor belongs to a specific test, channel and attribution setting. The definition of iROAS stays the same either way.
Why does it matter?
Incrementality asks what would have happened without the activity. iROAS puts a number on the answer. It also connects to profit. Compare iROAS with break-even ROAS and you see whether the ads paid for themselves, not just whether they were credited. For why the two numbers differ, read platform reported ROAS vs incremental ROAS.
For ZFD: We want to see the iROAS next to the margin. Credit in the panel is not the bank balance.
Worked example
Easily confused with
iROAS vs ROAS
Credited revenue over spend, as the platform counts it.
iROAS vs Marginal ROAS
The return on the next euro of spend, not the average. Some definitions of iROAS are marginal ones, read from a response curve1. We keep the two terms apart.
Buzzword alert
"Our incremental ROAS is strong"
Usually means: a number called incremental, with the method left out.
Ask: what method sits behind the word "incremental"?
"We apply an incrementality adjustment to your ROAS"
Usually means: the platform figure multiplied by a factor from some earlier analysis.
Ask: when was the factor last measured?
FAQ
Is iROAS always lower than platform ROAS?
Not necessarily. Do not assume the direction. Only a measurement tells you whether platform ROAS was too high or too low.
Can iROAS be calculated without a test?
Not on our definition. It needs an estimate of revenue caused by the ads, which comes from a test or a calibrated model.
Is a single iROAS number enough?
No. A number from one test is a range with a date on it. Ask how wide the range is and how old the test is.
Let’s find what’s actually working.
Next step. Compare credited and caused revenue on explicit assumptions. Use the Is your ROAS real? calculator.