When I’m talking to prospective clients, one of the questions I’m regularly asked is why they should invest in expensive predictive analytics software when they already have Excel. Excel has some useful features and for many organisations it can be a useful toe in the water of analytics but it’s really not built for the rigors of true predictive analytics. Organisations that stick to using Excel will never be able to truly reap the potential rewards that predictive analytics can offer. I’ve recently been working with a large B2B online publishing organisation and my experiences there illustrate this point perfectly.
The company wanted to review how it was using analytics. Its aim was to improve customer retention at the lowest cost and improve cross and upsell revenues to existing clients – B2B customers who generally bought structured publishing packages with 28 day cancellation terms. The sales team had separate outbound and inbound teams as well as an independent retention team who focused solely on cancellations.
The retention team were purely reactive, waiting until a client cancelled before offering discounts or package offers to tempt them to reverse their decision. This was successful in terms of the number of clients who were retained, but it was expensive. Sharp customers got wise to this practice and began to exploit it, knowing that if they threatened to cancel they would be offered a better deal. The inbound and outbound teams were separately responsible for new customer acquisition and for the sales of additional packages or upgrades to existing customers. There was no formal prioritisation of whom to target or with what offer.
The organisation had just invested in a team of highly qualified and capable analysts, one of whom built a detailed model for customer retention using Excel, which had taken around eight weeks to develop. This model estimated the likelihood that each customer would cancel as well as suggesting the offer most likely to successfully keep particular customers. The retention sales team had just started to use this model to inform their activities, enabling them to be more proactive in the way that they handled their customers.
The model flagged existing customers at most risk of cancelling, who were then called so that the retention team could explain the options in their existing package, the goal being to have them use what they already had more effectively. If a cancellation was received then the retention team would call the client and try to persuade them not to cancel by offering them whatever the model suggested would be the most appropriate and cost effective package. Some customers were flagged as low value and hence not worth retaining.
In practice, however, the Excel model was extremely difficult to use, hard to update and almost impossible to deploy to the sales teams. The file created was massive, cumbersome and easily ‘broken’ if used incorrectly. The model also had to be manually updated at regular intervals in order to keep it relevant and valid – this updating process took huge amounts of time away from the analysts. As the complexity of the process grew so did the number of models generated. The amount of data expanded to such an extent that the task could no longer be feasibly be completed in Excel.
The company came to Smart Vision because it wanted to find out if there were more efficient ways to create, maintain and deploy such models whilst also reviewing how to build a similar set of cross and upselling models without sacrificing the investment of time and resource already made. One of its analysts worked alongside one of our experts to rebuild the cumbersome Excel model using IBM SPSS Modeler. Together they were able to build a more effective predictive model for retention in less than two days, a massive reduction from the eight week build-time of the previous model.
The new IBM SPSS models went live in January 2013. These were deployed into the various sales teams to inform proactive campaigns as well as to determine reactive strategies. The result is that in less than a year the company has already seen a 20% increase in customer retention. This translates into savings that run into millions of pounds, covering the initial cost of the investment in IBM SPSS Modeler many times over. The new models are easy to deploy, simple to update and much more efficient to run. The company saves both time and money every time it deploys them.
I’ve seen results like this many times before. Companies may worry about the cost of investing in predictive analytics but the fact is that storing and analysing data in Excel is almost always a false economy. Investing into proper predictive analytics software such as IBM SPSS products enables faster, more efficient and more accurate models to be produced. This represents a huge saving in both time and money, and that’s before you even begin to consider associated savings that come from gaining higher quality predictions about your business.