What is a Control Group
Q: I’ve heard the term “control group”. Can you tell me what this means?
A: Gladly! A control group is a vital part of the direct marketing test process, because it helps validate testing results and helps to prove the ROI of a specific campaign. Simply stated, a control group is a test cell of customers or prospects who receive no special treatment. Imagine that you are trying to prove the concept that a catalog specifically targeted to your customer segment of families with children will boost sales of those types of products. So, you’ll identify which of your customers have children, and send the special catalog version to this group. In this case, a control group could be another segment of families with children who simply receives the normal, non-versioned catalog. If your versioning test is successful, the group that received the special catalog will respond at a higher rate, and buy more products than the similar, “control” group that didn’t receive the targeted catalog.
Another example of the use of a control group is when you are trying to validate your statistical model. For one of our clients, for example, we mail a modeled compiled list in some very targeted geography. For each mailing, in addition to selecting prospects who are most likely to become buyers (based on our model), we also select a random group of 10,000 names, selected regardless of model score. The high-scoring model should perform better than the control group, which validates that the model is still predicting what it was built to predict (propensity to buy). In other words, we should see more buyers coming from the high scoring group than from the control group. If we start to see the control group performing as well as the high-scoring group, we’ll know it’s time to re-build the model!
In a nutshell, direct marketers use control groups to validate the results of a variety of tests, from lists, to creative, to offer.