What is a baseline in marketing?
How does a baseline work?
You cannot see sales that did not happen because of an ad. You can only estimate them. The baseline is that estimate: the sales you expect if you did nothing different.
Add the activity. Measure the result. Subtract the baseline. What is left is the extra effect.
That is why the baseline matters more than the headline.
How is a baseline set?
There are a few common ways:
- A control group. A comparable group that does not see the ads. Its results are the baseline. A holdout test works this way.
- Control regions. Regions where nothing changes show what would have happened elsewhere. That is a geo test.
- The past. Last month or last year, adjusted for things that changed. Usually the weakest of the three, because seasonality, promotions and pricing all move sales on their own.
A control group is the stronger design because it faces the same calendar as the exposed group. A report that compares this week to last week has to explain everything else that changed.
Worked example
Easily confused with
Baseline vs Average performance
An average describes the past. A baseline estimates a counterfactual, meaning what would have happened without the activity.
Baseline vs Lift
Lift is the measured gap between exposed and control. The baseline is the control side of that gap.
Baseline vs Business as usual
A loose name for the same idea. It is only useful if someone writes down what "usual" includes.
FAQ
Can a baseline be wrong?
Yes. It is an estimate. If it ignores seasonality, promotions or price changes, any lift measured against it is off too.
Is a baseline the same as a control group?
Not quite. A control group is one way to estimate the baseline. The baseline is the number; the control group is how you got it.
Let’s find what’s actually working.
Next step. Check what your ROAS looks like on explicit assumptions. Use the Is your ROAS real? calculator.