INCREMENTALITY CALCULATOR
Incrementality & iROAS Calculator
Turn a holdout test into your true incremental ROAS, and see how much of your reported return the platform is taking credit for but did not create.
New to lift tests? Read the real thing: incrementality and lift tests first.
Take it further
Designing a real lift test?
Go through the design with a former Meta growth lead who has run hundreds of them. 30 minutes, no pitch, no retainer.
How to use this calculator
Run a holdout test, then enter the results. A holdout test splits your audience into two random groups: a control group that sees no ads, and an exposed group that does. Enter each group's size and conversions, plus the spend and the revenue per conversion, and the calculator returns your true incremental ROAS, the ROAS the platform would report, the incrementality percentage, and the number of conversions your ads actually caused.
How incrementality and iROAS are calculated
The control group tells you the conversion rate that happens anyway, with no ads. The exposed group's extra conversions, above that baseline, are the ones the ads caused.
Reported ROAS, by contrast, counts every exposed conversion as if the ad caused it, which is what the platform does. The gap between reported ROAS and iROAS is the value the platform is claiming but did not create.
A worked example
Say your control group of 10,000 people converts at 2% (200 sales) and your exposed group of 10,000 converts at 3% (300 sales), on €2,000 of spend at a €50 order value. The platform reports 300 sales, or €15,000, a 7.5x ROAS. But only 100 of those sales were incremental, the other 200 would have happened anyway. So your true incremental ROAS is €5,000 over €2,000, or 2.5x, and incrementality is 33%. The platform is taking credit for a return three times better than the ads actually delivered.
What is a good incrementality percentage?
It depends entirely on the tactic. Cold prospecting is usually highly incremental, often 60 to 100%, because it reaches people who did not know you existed. Retargeting and branded search are frequently only 10 to 30% incremental, because they mostly reach people already on their way to buy. There is no universal target. The point is to measure it, then move budget toward the spend that is actually causing sales. For the full method, see incrementality and lift tests in the measurement guide.
Is the result statistically significant?
This calculator measures the size of the lift you observed, not whether that lift is reliable. A small test can show a big lift that is really just noise. Before you trust a result, confirm the test was large enough with the test duration and sample-size calculator, run it for at least one full purchase cycle, and never stop early or read it the moment it looks good.
Common questions
Incremental ROAS is revenue divided by spend, counting only the sales the ads actually caused, not the ones that would have happened anyway. You measure it with a holdout test where a control group sees no ads and an exposed group does. iROAS is almost always lower than the ROAS the platform reports.
Incrementality is the share of conversions an ad actually caused, on top of the conversions that would have happened without it. If your exposed group converts at 3% and your holdout at 2%, one third of the exposed conversions were incremental and two thirds would have happened anyway.
Because the platform credits itself for conversions that would have happened without the ad, especially on retargeting and branded audiences that mostly reach people already going to buy. A holdout removes those, so the incremental figure is lower. The gap is the value the platform takes credit for but did not create.
Randomly split your audience into a control group that sees no ads and an exposed group that does, run for at least one full purchase cycle, then compare conversion rates. Meta Conversion Lift automates this, and a geo test where you change spend in some regions and not others is harder to game. Size the test first so it can detect the effect you care about.
It depends on the tactic. Cold prospecting is usually 60 to 100% incremental because it reaches new people. Retargeting and branded search are often only 10 to 30% incremental because they mostly reach people already on their way to buy. Measure it rather than assume, then reallocate toward the incremental spend.
No. It measures the size of the lift you observed, not whether it is reliable. A small test can show a large lift that is really just noise. Confirm the test was large enough with a sample-size calculator, and do not stop the test early or read it the moment it looks good.
Related: the paid social measurement guide, the test duration calculator, and the break-even calculator.