For certain propensities it can be useful to conduct experiments to check that the predictions are working the way they should, and to confirm that the action you decide to take has the desired effect. A good example here is regular giver churn. You might want to understand if the donors that we predict are going to churn actually do churn, but you might also want to test if the action that you took, such as giving them a thank you phone call, resulted in a reduction in churn.
In order to conduct an A/B test you need to create two statistically matched groups. Dataro can provide you with custom assistance to generate those groups using random allocation, and we can assist you to track the progress of both groups through the experiment window. Using the churn example, that experiment window would normally be a period of six months.
Comments
0 comments
Please sign in to leave a comment.