Every database promotion should include a control group of customers that are not exposed to the actual promotion. The success of the promotion is measured by the differences in response of the promoted group compared to the control group. When conducting an experiment, a control group is the basic group which does not receive any treatment or neutral treatment. To see the results of the experiment, the person who is conducting the experiment matches the results between the treatment group and the control group. When there is a new product which needs to be launched to the public, it goes through a control group to see how people will react to it. So, when there is a new version of a cell phone being done, before sending it to the stores, and making a promotion of it, there should be a group of people controling it, seeing if it works, how it works, does it have some bugs or mistakes in it. If everything is OK, than it is launched to the public, if not it is returned to the factory to correct the mistakes. This way time and money are preserved. It is easier and cheaper to pay a group of people to test the product, rather than to pay for repairing the item, every time a customer brings it to the shop.