Data Mining and the future of Retention
Data mining is important because we can use data in predictive analytics and as an industry we can use it to help our customers achieve more from their goals but also just stick to exercise more generally.
It also helps us understand what’s going on within our business. So we can plan more efficiently and introduce products that have a bigger uptake and a bigger outcome in the end for the customer.
This is where digital retention will be going. Not just in what I do but in what other people are doing as well. We data mine the membership database.
This is why the data becomes really important.
We can create what's called a persona. For example, I am a 52 year old male. Does that mean all 52 year old males are like me? No, some 52 year old males like smoking, drinking and watching football. I like watching football, but I don't smoke or drink so our personas are different. The types of sports I like to be engaged in or activities I'd like to be engaged in are things like the gym, but also mountain biking.
A persona can start to group and identify common characteristics, not of what we do, but how we behave and what turns us on.
When they profile me, one of mine is exhilaration, which comes out in the mountain biking. So if they've got programs, ‘we've got this on offer’, it’s going to look for things that would be exhilarating in the workouts.
If I'm more of the calm laid back type it's going to say, ‘we've got this new yoga program’, so it's not going to do it by age. It's going to be done by the types of things that are going to turn me on.
Facebook did a study of this with Cambridge university in the UK, they managed to get it down to one question.
Answer this one question and we can tell you what your persona is. We can classify you in different ways. We can do time series, which is the traditional retention analysis.
We can look for associations between what I do, what I like and what other people are like. So you can then create groups.
We can do regression analysis, which means we can test what the relationship is between exercise intensity and exercise adherence, and we know, this is just a general example, that as exercise intensity goes up, exercise adherence actually goes down, which should mean we start to question giving HIIT to everybody.
We can segment, we can do lookalikes. This is really good for marketing and advertising. Look at the profile of your members and then look at the profile of your community and then identify if you’ve got some more of that type of member out there and go get them.
And we know what the persona of these people are like, so we can encourage people to write our marketing and do our follow-ups in the same way.
This wasn't designed for retention, it was designed around sales, but I looked at it and said, you could flip that and turn it into retention, because you can start collecting data on somebody and this could be all the different data streams about someone.
There could be up to 600 columns. We scraped Facebook. We scraped Twitter. If you join us on Facebook, we can get your data. If you follow our Twitter account, we can get Twitter information about you.
If we've got your postcode, we can get about 40 pieces of data about you and we use that to create a model based on you, not someone else, one for you and another one for someone else and another and another, so we can look at your unique behaviours within the business.
That allows us to update our knowledge about you and the types of ways in which we can support you, interact with you and help you be successful at exercise.
That data stream continues to feed January, February, March. So everything you do, everything you feed, just updates my knowledge further and further about you and I can get more and more accurate in the way that I feed you information.
A bit like Amazon does. If you shop on Amazon, a bit like Netflix does, if you watch Netflix. So these are the same processes they use.
