• Split Testing Should Be Reductive Not Binary Nov 20, 2020
    Views: 187

    One of the few benefits of lockdown is that every week my family enjoys what amounts to an online get-together where we catch up on what’s happened to the various members. We also play together, despite restrictions. One popular party game is what we call 20 Questions, although it probably has any number of other titles. I’m good at it, but then I know email marketing, so should be.

    We choose an object from a restricted list; for instance, ‘something in a garden’. The remaining four sides of the family ask a yes/no questions each in rotation and after 20 make a guess as to what the object is. Other rules have evolved over time.

    We’ve been playing this game every Sunday for over six months and you might think we would be getting bored with it by now but, if anything, we hurry through the, rather repetitive, gossip period because we have so much fun playing the game. It’s an intellectual exercise. We get limited information and, working from that, reduce the variations until we can classify the object. It’s just like interpreting data on our email marketing lists.

    It’s been mentioned before by many commentators but it’s worth repeating; a single split test of an email marketing list is, essentially, binary, just like our game. The only answers we receive are better or not. From that simple base we have to work out the preferences of our subscribers in order to ensure that we send them offers on products they need.

    When working out what you will question in your next campaign with split email marketing lists, think of not only that specific question, but the one you asked before, and the ones before that. You should be honing in on a specific answer. Just like our 20 Questions.

    That changes the system from binary to a complex series of questions which can define your subscribers to a considerable degree. Split testing should not be binary but reductive, in the same manner as our parlour game. Someone on Sunday was able to conclude that one object from the garden was ‘next door’s cat’.

You need to be logged in to comment