Thought Leaders in Big Data: Interview with Oliver Downs, SVP of Data Sciences, Globys (Part 3)

Sramana Mitra: That concludes one use case. Why don’t we do a few more?

Oliver Downs: Another scenario – same data but different KTI – is an operator in a developing rather than a developed market. You can imagine there are users who really aren’t that engaged with the service but they are keen to have access to a mobile device, which they perhaps wouldn’t use very frequently. Even though she is in a developing market, they behave somewhat like my mother-in-law. When I text message her a picture of my children, she will call me back from her land line to tell me how great the picture was.

You can identify these groups of users. The chance could be is that you can actually see, viscerally, the fact that they are on the wrong type of rate [and] time structure for the way they behave. Customers recharge, and they have a balance that will roll over when they recharge next. But if they don’t recharge within a certain period, they will lose their balance. The problem is that they are such light users of the device, they never consume a significant proportion of the balance they are carrying.

Again, we can identify that there is a large group of customers who have what we call step-up behavior. They are continually recharging to maintain their balances. That makes them high average revenue per month users. But in fact, their consumption doesn’t at all reflect how they have to spend to maintain the service. The interesting thing about that is these customers aren’t perfect. Sometimes they miss the timing for their rollover, and it can cause them to lose a very large amount of balance – perhaps they rolled it over for several months in a row. Of course the challenge is, when you notice someone who wants to maintain the ability to use his or her device, your immediate instinct is not, “I am going to cancel my service and go to someone else” but, “I still need my device, I better recharge immediately and start from the beginning.”

Naively, and particularly without this kind of sequential view of customer behavior, you might imagine that this big loss of balance event actually triggers the customer leaving the mobile operator, when in fact the most common event following a big loss of balance is to recharge again. It is 71% more likely that a customer turns to subsequent recharge cycles down the road – when he or she is still reflecting on what has happened, and that person had time to put in place another mobile operator for contingencies.

When you can observe the combination of recharge behavior, usage, and consumption behavior, it is very clear to see that while you are maintaining them with a very high average monthly revenue, your long-term attention prospects for that customer are very poor. As a result, you can tweak a rate plan offering that might decrease the monthly average revenue for that user, but might keep that customer for you in the longer term. We learned that churn is not an immediate byproduct of this loss of balance event, but a consequence several steps down the line. Being able to act on it has a significant impact on customer retention.

SM: What kind of ROI do you deliver through this process?

OD: This is a 10% to 20% improvement in terms of active customer base retention.

This segment is part 3 in the series : Thought Leaders in Big Data: Interview with Oliver Downs, SVP of Data Sciences, Globys
1 2 3 4 5