An incrementality test is an experiment that measures the sales an ad investment produced that would not have happened without it. It shows the gap between the conversions the panel reports and the real contribution. Part of the audience is held out from the ads, the two groups' sales are compared, and the difference is the net return of the advertising. It is the only approach that ends attribution model arguments.
The most common implementation is a geo split: campaigns are switched off in one of two comparable groups of provinces and total revenue is compared over two or three weeks. Sufficient volume is essential; at a few orders a day the result does not separate from noise. For most brands the result comes in below the panel number, especially on brand search and retargeting. That is why what will be done with the outcome should be written down before the test; otherwise a bad result gets talked away and the budget stays where it is.
Incrementality = (Test group sales − Control group sales) ÷ Control group sales
With 100 units of sales in the provinces running ads and 82 in the provinces switched off, incrementality is 22%; the panel may report far higher contribution over the same period.
The most common implementation is a geo split: campaigns are switched off in one of two comparable groups of provinces and total revenue is compared over two or three weeks. Sufficient volume is essential; at a few orders a day the result does not separate from noise. For most brands the result comes in below the panel…
Knowing the term is not enough; you need to know which lever to pull. In a free 30-minute call we look at your store's actual numbers.
Get a free analysis