Better science produces more profitable models
Propensity models help you to maximise Return on Investment (ROI) by targeting the most suitable audience. Fundamentally, propensity models allow you to score customers or prospects according to their likelihood to undertake a given action, for example; likelihood to become a customer, likelihood to buy again, likelihood to lapse or likelihood to take up a new service.
Qbase models are multi-variate, meaning that our models are based on a multitude of factors, including demographic, transactional and environmental. The models we build use a range of statistical techniques to measure the impact of combining these variables together.
"This is a significant improvement over the uni-variate models produced by most of our competitors". Uni-variate models measure the impact of each unique variable and so cannot identify the impact of combined factors.
Look at the following basic example. A mailing campaign produces a response rate of 4.5%. The result of a uni-variate analysis is shown below. It enables us to identify the impact of 4 key factors. Clearly, customers who previously ordered in the past 30 days show the highest response at 8%.
However, the multi-variate analysis above identifies the impact of combining the factors, (the table shows just a few of the total permutations). Combining the factors shows us the highest responders generated an 11% response – the model is more "discrete".
Qbase propensity modelling enables us to score each customer or prospect, based on the permutations of their individual factors and the importance of each of those factors. Then you can improve your targeting and invest a larger proportion of your budget into the higher-performing segments. The end result is increased return on investment.