Behind all the news headlines about credit losses, write downs and foreclosure there are collections departments working overtime. Not only is recovery in current times difficult, but it’s also expensive. Collections are the highest operational cost once an account is on the books with 80% of the operational cost of a portfolio attributable to collections costs and write offs.

While a small improvement in recovery results can make a significant impact on operations, collections strategy challenges abound – who, when and how should businesses focus their collections efforts in order to maximize recovery, minimize costs, and maintain (or improve) customer satisfaction?

In traditional collections operation, activities were based on outstanding balance, days past due, and credit score. Early stages of delinquency brought less activity with less experienced agents, while in later stages, more experienced agents took on more aggressive tactics. Not only does this model not scale well to address situations like we see in the current climate, it’s also inefficient. Many of the early stage accounts would likely have paid without any intervention whatsoever.

This is where applying predictive analytics to segment debtors becomes critical to triage the AR portfolio. Those we can segment and predict as ‘sloppy payers’ – customers that simply pay late - can be ignored as they typically pay without any action taken at all. Those we segment as under economic hardship are likely to pay if sufficient action is taken. And fraudsters, they won’t pay regardless.

Once we have segmented our portfolio, we can define and assign strategies and treatments. In early stages of delinquencies, for example, there are large numbers of ‘sloppy payers’ and here we could apply a strategy where we target those least likely to pay (i.e. those under economic hardship). Sloppy payers will pay anyway (so it’s a waste of resources to focus on them), and in fact may suffer a hit in client satisfaction if annoyed by collection efforts. In later stages, debtor segmentation would help us target those most likely to pay – collections are a competitive environment and late-stage debtors likely have other creditors.

While all of this sounds simple, there are some key areas of analysis required in order to get good predictive results in our debtor segmentation. We’ll pick this up in our next post.

Your customers and prospects are predictable. They engage, inquire, negotiate, purchase, and yes, sometimes even attrite, according to quiet little patterns etched into transactional and behavioral histories stored in your very own CRM system. Organizations focus predominantly on using this data to foster insight into understanding “what happened” or “what is happening”. And while applying a BI strategy to grasp ‘rear-view’ sales and revenue trends is important, the same pool of customer data remains a largely untapped source for valuable predictive insight.

There are two basic business challenges in the sales channel that operational CRM systems cannot fully address, as effective as they are at managing the orchestrations around our client contacts. First, there is the cost of sales itself, or your net sales expense. Sales representatives need to build pipelines, develop accounts, and mine leads in order to ultimately close business. These activities, as well as the efforts of those that support this process, are all costly. Sales and business leadership remains under increasing pressure to meet targets without increasing cost, which can only be accomplished through further improvements in productivity.

Secondly, sales teams - and the sales representatives and account managers that comprise them - are given to very human foibles, and not least among these is the emotional subtext that enters into the sales process. The purchasing intentions of clients and prospects are mediated and interpreted by sales who in turn apply the sum of their experience and intuition to forecast outcomes of these engagements. As a result, the aggregate forecast for which the sales executive is ultimately accountable is only as reliable as the individual contributions of those very human sales reps and account managers themselves.

Both these issues are fundamental – how can organizations become dispassionate in their interpretation of buyer intentions, and how further can sales productivity improve?

Given that wealth of untapped sales and customer data, this is where predictive analytics comes in. Consider some of the inefficiencies in the demand generation process. It is common for organizations to maintain thousands or even tens of thousands of sales leads in their systems and yet statistically only a very small percentage of these ever convert. One relevant application of predictive analytics is the indication of which leads from a database of several thousand actually matter, such that sales teams are able to bolster productivity by converting more leads (and hopefully closing more business) with dramatically fewer calls. Similarly, consider the scenario of the sales manager looking at a pool of open opportunities and a looming quarter end. The most productive use of resources would be one where the sales leader could ensure coverage of the opportunities most likely to close by the end of the quarter, another application of predictive analytics.

Predictive analytics can also address the human issues inherent in forecast reliability. Individual interpretations of buying propensity give way to a scientific extrapolation of forecast results that not only applies data mining and statistics to improve accuracy, but also helps to eliminate the surprise factor that impacts sales teams at the end of fiscal periods often regardless of their maturity. Many organizations also use predictive forecasts (and its delta with an individual’s forecast) as a coaching tool to improve sales forecasting over time at the individual level and team level.

The breadth of applications for predictive analytics for sales and marketing are many. With the maturity of CRM systems and the wealth in client data they are silently accruing, the convergence of these technologies is not only complimentary, but imperative.