There are great events you can invite yourself to, and I am not talking about some kind of Facebook flash mob party here, but about serious business. For example think about our forthcoming Teradata’s Partners Conference with customers and IT professionals from leading companies all over the world gathering from October 21st to 25th in Washington DC.
At other events in life, access is far more restricted, meaning you can’t simply register and pay your fee in order to attend, especially in private life. Speaking of which, has it ever happened to you that one of your supposedly best friends didn’t invite you to a turning-point event in his life, e.g. his wedding? If so, you most probably reflected heavily on all the conversations you had with him over the last years. Where did things go wrong? But what's worse: simply writing him a nice e-mail to make amends won’t do. You most likely and sadly lost a friend – for good. The business equivalent to not being invited to a wedding is not being on the shortlist for an important bid.
However, the good news is, that in today’s business you have means to analyse these relationships (which often count by millions) at an unprecedented level of precision which helps you to understand the ups and downs. By deploying some sophisticated data modelling, this gives you quite a reliable outlook to the future. More on this below. But let’s stick to the smaller circle of our wedding example for the moment: Our networks of interaction are not that sophisticated. We are regularly in contact with family members and relatives, friends and colleagues. As the British psychologist Robin Dunbar found out, the design of the human brain limits the number of our important contacts to approximately 150 stable social relationships. In the wedding example, you obviously couldn’t manage to stay in close enough contact with this friend and the relationship deteriorated over time. Maybe he simply felt he was number 151 on your list and he decided that this rank didn’t warrant an invitation.
Back to business: Just as in any other relationship, the one between a customer and a company deteriorates over quite some time before “the customer turns away”, as we previously described on this blog. Such deteriorations might be based on insufficient service, or the failure not to make a customized offer before the contract expired. Uncovering the underlying mechanisms is crucial in order to reduce customer churn. The models I referred to above can tell you the right time to contact a customer - a highly valuable information, as most of my peers in the industry see it. Now imagine my surprise when I read that “predictive analytics is utterly useless” in a comment by Max Pucher on Forrester analyst Mike Gualtieri’s blog. While Mike Gualtieri already collected some strong arguments to make the case for predictive analytics, Pucher believes that “the complexity of interactions cannot be causally analysed and thus not be predicted”. So, let's have a look:
The telco sector is one of the most advanced in the area of predictive modelling and analysis. The sole number of our customers applying such methods in this industry, where all of the top ten companies worldwide rely on Teradata, easily refutes Max Pucher’s statement. Take, for example, Telefónica Czech, which uses Micro Targeting, texting the customers who are most likely to visit a store. Or take the Russia-based Telco MTS, that uses Social Network Analysis to predict customer churn and apply appropriate measures before the customer takes the final decision.
How about examples from other industries? With a dedicated analytical CRM (ACRM) system, which also leverages predictive analytics, Teradata customer BAWAG Bank, one of Top 5 Austrian financial institutes, has had considerable success, too. Branch advisers who used model-generated leads achieved 12% success rates in their cross- and up-selling activities compared to rates of between 1% and 2% in a control group.
Really, “no business benefits”? I think, Max should seriously think about getting onboard a plane to the Teradata Partners Conference and join some of our sessions there, which might be the Road to Damascus for him with regard to predictive analytics. I would highly recommend, for example, the session of Peter Huang, Sr. Manager BI Architecture and Mikael Weigelt, Director Customer Relationship Marketing at T-Mobile. They will tell us how the mobile operator uses statistical models to predict customer churn, purchase propensities, fraud, or payment behaviour with a focus on the automation of such processes. Doesn’t this sound like an invitingly good offer? Registration is still open!