Incrementality Testing
StrategyDefinition
Incrementality testing measures whether your advertising produces results that would not have happened without it. It answers the question: "Did this campaign cause conversions, or would those customers have converted anyway?" This is the gold standard for measuring true advertising impact.
Standard attribution counts conversions that happen after an ad interaction, but many of those conversions would have occurred organically. Brand search campaigns are a classic example — users searching your brand name likely would have found you without the ad. Incrementality testing isolates the causal lift.
The most common method is a geo-based lift test: run ads in some regions (treatment) and pause them in matched regions (control). Compare conversion rates between groups. The difference is your incremental lift. Meta offers built-in conversion lift studies. Google offers geo experiments in Google Ads.
Design an experiment with treatment and control groups. For geo tests, select matched market pairs with similar characteristics. Run the test for 2–4 weeks minimum. Measure the conversion difference between groups. Calculate incremental CPA by dividing spend by the lift in conversions (treatment minus control).
Without incrementality testing, you cannot know if your campaigns drive growth or merely capture existing demand. Brands that test incrementality often discover that some high-ROAS campaigns have low incrementality (they were taking credit for organic conversions), while some low-ROAS campaigns drive significant new business.
Related Terms
Related Services
Frequently Asked Questions
-
Start with a geo-based test. Select two similar regions. Run ads in one (treatment) and pause in the other (control) for 3–4 weeks. Compare conversion rates. The difference is your incremental lift.
-
Start with your largest spend campaigns and branded search. These are most likely to have inflated attribution numbers. Testing brand campaigns often reveals the majority of conversions would happen organically.
-
Minimum 2 weeks, ideally 4 weeks. The test needs enough data for statistical significance. Longer tests reduce the impact of seasonal variance and give more reliable results.
Are your campaigns driving real growth?
We design and run incrementality tests that reveal which campaigns truly drive new business and which just take credit for existing demand.