What happened because of the activity?
120 − 100 = 20 additional purchases in the test group. This example does not calculate statistical confidence.
01Define treatment, control and the measurement before starting.
Why you can't trust attribution alone
Attribution distributes credit for sales that have already happened — it can't know whether the sale would have come without the ad. Retargeting and brand searches look fantastic precisely because they hit people who were on their way anyway. Incrementality asks the only question that matters: what would have happened without the effort?
The answer isn't found in Ads Manager, but in an experiment: hold something back, and measure the difference. That's the principle behind both geo-lift and holdout tests.
What happened because of the activity?
120 − 100 = 20 additional purchases in the test group. This example does not calculate statistical confidence.
01Define treatment, control and the measurement before starting.
How to run a geo-lift test
Split your markets or regions into two comparable groups: a test group where the effort runs, and a control group where it's turned off or held flat. Run it long enough for the difference to be statistically credible, typically a couple of weeks to a month depending on volume. The difference in total sales between the groups is the incremental effect.
The advantage is that you don't have to trust cookies or windows — you're comparing actual revenue in two groups. Meta's Conversion Lift and Google's geo experiments automate parts of it, but the principle is the same: hold back, and measure the difference.
What you typically find
Most brands running their first holdout test discover that some of their "best" attributed channels are far less incremental than the dashboard claimed, while the demand-generating top of funnel is more valuable than last-click gave it credit for. That changes how budget should be allocated.
From above, MER can give a cheap, ongoing check: if you raise budget and blended efficiency holds, the growth is incremental. Formal lift tests and MER complement each other: one precise and occasional, the other continuous and cheap.
Frequently asked questions
What's the difference between a geo-lift and a holdout test?
Both hold something back and measure the difference. Geo-lift uses geographic groups (regions/markets), while a holdout typically keeps a random share of users outside the effort. The principle — measure the added effect against a control group — is the same.
How long should an incrementality test run?
Long enough for the difference to be statistically credible — typically a couple of weeks to a month, depending on your volume. Too little data gives an uncertain result; too long costs unnecessarily in lost effort in the control group.
From insight to action
See how it applies in practice.
Website, channels and profit measurement
The Løgbutikken engagement covered the website, email, Meta, Google Ads and ProfitMetrics. The business grew and was subsequently acquired.
Read the caseMeasurement and customer economics
Discuss measurement and bidding
We review the values you measure and how they can inform budget decisions.
Choose a time30 minutes · No obligation
About the author
Nicklas Segatz Mortensen
Growth hacker and fractional CMO with 10+ years' experience and hundreds of millions in managed ad spend behind him. Background from larger Danish and international scale-ups, and from the agency world.
Meet the team