5th May, 2008
Stochastic Solutions has been announced as the winner of Hillstrom's Etail Analytics Challenge, a competition that required detailed analysis of two email campaigns to determine effectiveness and future targeting strategy. The competition attracted over 100 entries from around the world.
The submission from Stochastic Solutions utilised an approach based on Uplift Modelling approach that has been developed by Nicholas Radcliffe, first at his former Edinburgh start-up Quadstone Limited (now part of Portrait Software) and now at Stochatic Solutions.
Uplift Modelling is based on the fundamental realisation that classical direct marketing approaches to customer targeting are trying to solve the wrong problem. At it simplest, classical approaches try to predict who will purchase after some marketing activity, rather than whose behaviour is changed by that activity. By picking at this problem for over six years, Radcliffe has shown thatin some cases the targets identified by classical approaches are not merely suboptimal, but are sometimes people who are actually so turned off by the activity in question that they become less likely to purchase (so-called negative effects).
Over the last few years, leading banks and mobile operators have become increasingly attuned to this issue, especially when they find that heavy investments made in customer retention often backfire so badly that they end up driving more customers away than they save. A number of leading banks and telcos have, over recent years, trialled and then adopted uplift modelling, resulting in dramatically better targeted marketing activity, often achieving the holy grail of improving results and reducing spend.
Stochastic Solutions has now applied these methods Hillstrom's email marketing campaigns, resulting in an analysis that Hillstrom judged superior to all 120 other submissions.
The winning paper is available for download here.
A press release is also available here.