Calibrating Marketing Mix Models with lift tests
Written by Mikko Ervasti
Updated over a week ago

What are lift tests and why are they useful?

Lift test are carefully designed experiments where the objective is to find an accurate estimate for how a certain marketing activity drives incremental sales within a certain target audience.

Lift test results can be used to provide Marketing Mix Models more information about the behavior of a specific medium. It is helpful especially in situations where the signal-to-noise ratio is not optimal.

How do lift tests work?

In the lift test experiment, the target audience is split randomly to two groups: test group and control group. The control group doesn't see the ad campaign and thus they are the baseline for the experiment. For example, Facebook's conversion lift test shows an empty space in the area where the test group sees an ad. Test group sees the ads in the test campaign.

After the lift test is conducted, the results of the control group and test group are compared. From this one can calculate the following (note: illustrative example, real world needs a bit more statistics):

  • Incremental sales = Test group sales - Control group sales

  • ROAS = Incremental sales / Media investment

Below is an illustration of the Facebook conversion lift test (source:Meta) flow:

Did this answer your question?