In a meeting recently with the data team, we were discussing why seriously complex predictive models weren’t yielding the expected results and subsequent ROI.
One of the data guys who creates some amazing models for our clients then said something which stopped me in my tracks.
He felt that the models were too good.
If we predict that a certain portion of the audience are most likely to renew/upgrade etc, and we trust the predictive models why should we be surprised when our audience behave in the manner predicted – even before we have had a chance to communicate with them.
Simple, logical and somewhat counterintuitive to common marketing thinking.
Ultimately the client should be focusing their effort on convincing the people that aren’t likely to behave in the desired way. This resonates slightly with a point in one of my earlier blogs that we make the mistake of looking for people most like our customers instead of allowing customers to identify themselves.
A second, and in this case, a more important point is how do we get into a situation like that? My belief is that we tend to allow technology to dictate our approach somewhat. This immediately throws you into a tactical, executional mode.
We often see the communication vehicle being pushed ahead of thinking about how best to achieve our communication goals. Turn it around and think about what your customers are doing and thinking for a few minutes before jumping on that cool new technology train.
0 Responses to “Are we becoming too good?”