How it works
The Provider-Web
DBM Mailing-List



Ask Us!

The Tasks of Database-Marketing:

Definition of Control-Groups

In this chapter you will learn, how to set up control-groups and what the main usage of control-groups are.


Control-groups are selected customers or prospects you use for testing the impact of a specific activity. Typically control-groups do either NOT PARTICIPATE in this promotion or participate in a different activity-design, so that you may compare the strengths and weaknesses of each individual activity or design.

Selection of a control-group

You may select a control-group either randomly or representative for the whole selection.

Randomly means:
i.e. you chose every x-th customer from the total selection for your control-group

Representative means:
you choose the control-group that way that is represents your typical structure of the selection. i.e.:
if your selection consists of - let's say -
20% US-north
30% US-east
40% US-west
10% US-south
you would have to select the control-group that way, that it reflects the same percentage of the total:
20% US-north
30% US-east
40% US-west
10% US-south

Using Control-Groups To Identify Customers That Will Buy Even Without Promotion

When making a selection and afterwards promoting everyone of this selection, you typically have no way to tell, how someone (of this selection) would behave if you had NOT promoted him. This might be valuable information since perhaps this customer would have bought your product without any impulse from you. To find out which customer would have bought anyway, you take away a certain percentage of your selection (maybe about 5% of the total group) and do not promote these customers. This group we call "Control-Group".

When the promotion is over and it comes to measuring the results of it, you will analyze the results of both:

  • customers you selected and
  • the control-group

Typically you will find out that even customers that you did not promote still contributed with some sales. You may now start to analyze these type of customers to find out their characteristics. This information you may use when you plan and design your next promotion - meaning: you should not include this type of customer (since they will buy anyway) in your next selection. By not promoting them, you can save the money.

As you see, it sometimes makes more sense NOT to promote the customer!

Using Control-Groups For Testing The Impact Of Different Promotion-Designs

If you plan to launch a "big" promotion you will think of pre-testing different designs to identify the most successful one. And this one you will implement using your total selection of customers. The only question that still remains open is: which is the most successful design?

If you have a budget that is enough to employ a market-research company, you may skip the rest of this section ... If NOT, this section may save you quite a lot of money!

Database-Marketing can give you excellent support to find out which design should be preferred. The technique you use is again "Defining Control-Groups".

As described above you split your total selection file into some control-groups - for each of the designs to be tested you select one control-group. But to ensure more accurate testing results you should use representative selection (so that the control-groups are a typical representation of the total selection file and that all of the control-groups look the same). The reason for this is that otherwise you might face some serious problems with comparing the result since they come from a totally inhomogeneous group.

Now you may start testing each of the designs by sending it to a control-group. When finished, analyze the results generated from this control-group. And now: compare the different results from the different designs. All that is left now, is that you identify the design that fulfilled your tasks best (whatever your tasks were ...).

By applying a promotion-design to only a small target-group, you can easily identify whether it works or not. Tracking the results and afterwards comparing them is easy work if you have implemented Database-Marketing.

This WWW presence, the layout and any pictures used in this site is © copyright protected and powered by
G.H.E.M. Global Home of Electronic Marketing - 1996-2011.