Incrementality

Incrementality is the measure of how many conversions or how much revenue your advertising actually caused, beyond what would have happened anyway. Platform reporting shows conversions that followed an ad within an attribution window, but some of those people would have bought regardless. Incrementality tries to separate cause from correlation.

How it is measured

The core idea is a controlled comparison:

• Test group: people who can see your ads.
• Control group: a comparable group held out from seeing them.

The difference in conversions between the groups is the incremental lift.

Worked example (made-up numbers): in a holdout test, the exposed group produces 1,000 purchases and an equally sized control group produces 700. The incremental purchases are 300, so 30% of the exposed group's purchases were caused by the ads.

Common methods

• Conversion lift studies: at the time of writing, Meta offers lift testing for eligible advertisers, which randomly holds out part of the audience.
• Geo tests: turning ads on or off in some regions and comparing them with similar regions.
• On and off tests: pausing a channel for a period, which is simpler but more affected by seasonality.
• Marketing mix modelling: statistical models of spend and sales over time.

Why it matters

• Retargeting and branded search can look excellent in platform reports but show lower incrementality.
• Prospecting may be undervalued by last-click reporting.
• Incrementality helps you decide where the next unit of budget should go, alongside MER and ROAS.

Tests need enough volume to be reliable, so smaller brands often start with simpler geo or on and off tests.