Analytics-driven solution helps sales teams identify opportunities

In collaboration with the Enterprise On Demand Transformation group, IBM Research has delivered two tools – OnTARGET and MAP – that use predictive analytics to better identify new IBM sales opportunities as well as effectively deploy sales resources to IBM accounts with the best forward-looking revenue opportunity.

Sales teams in nearly every major company face two key challenges: First, how can individual sales representatives identify new sales opportunities at existing and prospective client accounts? Second, given that sales resources generally go toward accounts that already drive in revenue, when should sales reps focus their efforts on accounts that are likely to generate revenue at some point in the future?

To help sellers find revenue growth opportunities at existing and new accounts, IBM researchers have developed two Web-based tools: (1) OnTARGET, which predicts the likelihood that a company will buy within an IBM product group; and (2) MAP, which estimates the true revenue potential at each account within each major IBM product group.

The models for these selling tools grew out of an internal initiative called the Market Alignment Program in which researchers validated the model-estimated revenue opportunities by conducting extensive interviews with front-line sales teams. The resulting estimates were then factored into a process whereby sales resources were allocated to IBM client accounts with the best anticipated revenue potential.

Borrowing strength between modeling problems

OnTARGET has spawned innovative approaches for handling limited data in predictive modeling technologies that IBM researchers call “data-enhanced modeling.” The main idea here is to combine modeling problems where data is sparse, “borrowing strength” between modeling problems and carefully managing the bias-variance tradeoff involved in these decisions.

The “customer wallet” and opportunity estimation models developed for MAP implement two approaches to a critical business problem. One approach is based on defining “realistic wallet” as a high quantile of the conditional distribution of customer spending and adapting quantile modeling approaches, such as quantile regression, for this task. The other approach addresses a more standard definition of “served wallet” as the customer’s potential budget to spend with IBM. This leads to a latent-variable graphical modeling approach, which allows researchers to make inferences based on this quantity (served wallet), which, of course, IBM researchers would not be privy to.



Web-based tools are already in use

Some 7,000 IBM sales professionals in 21 countries across three major geographies are now using the OnTARGET Web-based tool. Analysis of IBM sales closed after deploying the tool demonstrates that the models are assigning a high propensity score to sales opportunities that ultimately bring in revenue. Meanwhile, the MAP tool has helped conduct hundreds of interviews with sales leadership teams across all IBM geographies and it continues to play an important role in the deployment of IBM’s sales resources.

During the 2006 deployment, for example, global sales teams participated in approximately 420 Market Alignment Program workshops. Based on the resulting interviews, several hundred sales resources were moved to higher revenue-opportunity “invest” accounts. Examination of these accounts has revealed a 19% increase in sales-opportunity pipeline relative to the prior year. Quota attainment among these redeployed sales resources was significantly higher than it was when sales resources were shifted as a result of other initiatives.