The science behind decisioning: strategy optimisation
Every day, billions of decisions are made about customers and prospects: which offer to make; which limit to set; which product to cross-sell; which collections action to take. But just how much science lies behind those decisions? Many organisations collect enough data to build scorecards and propensity models. In doing so they are taking a step towards scientific decisioning.
With one model and one objective, decisioning is a relatively straightforward process. A risk scorecard predicts customers’ probability to default and setting a score cut-off (a point at which customers are declined) has a predictable effect on default rates. Likewise, a marketing propensity model predicts customers’ probability to respond and setting a score cut-off will have a predictable effect on sales.
But in the real world, performance indicators such as default rate and sales are only part of the bigger picture. If decision strategies are to be truly strategic, they need to consider wider goals and objectives: expected loss, lifetime value, provision for bad debt, return on capital, net present value, exposure and much more. Regardless of which metrics are considered, the common factor is that a decision has multiple dimensions of consequence. And with each dimension, the decisioning process becomes more complex.
''If decision strategies are to be truly strategic, they need to consider wider goals and objectives..''
The good news is that there is a significant reward awaiting the organisation that applies a scientific approach to decisions. It becomes a source of competitive advantage; a classic example of this is profit-based pricing. If an organisation offers low-risk prospective customers exactly the correct incentive to take up an offer, they will do so and add to the quality of its customer portfolio. If high-risk customers are asked for the correct premium, they are likely to take up a competitor’s offer who doesn’t apply risk-based pricing, reducing the quality of the competitor’s portfolio.
Modelling the consequences of decisions is big step forward, but applying those models to the greatest effect is a challenge. It doesn’t take a great deal of complexity before a traditional decisioning system based on business rules starts to leave value on the table.
A classic example is whether to cross-sell product A or product B to a list of customers. A response model exists for both offers, so each customer should be offered the product for which they have the highest probability of response. Easy so far, but what if there is an objective to sell at least 500 of product A, and the first strategy only delivers 300? The problem is now much more complex. Only one further constrained dimension, such as the minimum expected value generated by the sales, and the problem is practically impossible to solve effectively with traditional techniques. Value is left on the table and there is a need for more science behind the decisioning.
This is just one example. Data-driven decisions are made across the whole of the customer lifecycle from prospect marketing, through originations, customer management, retention, collections and charge-off. In many cases, there are operational or strategic constraints within which to work, there are conflicting performance indicators to maximise and there is a database recording previous decisions upon which expected outcomes can be predicted. Each such case is an opportunity for optimisation.
Strategy optimisation enables an organisation to assess the effect of different actions, decisions, limits or terms on profit and other business metrics. It produces clear information on the trade-off between different decision scenarios so the business can understand the effect of different constraints on business profitability. Optimisation then mathematically identifies the optimal mix of customer decisions and actions that maximises the value of the overall set of customer interactions, within the limits established by the constraints.
As competition grows across global marketplaces, organisations will come under increasing pressure to make effective decisions. Those that understand and exploit the science behind decisioning will have a competitive advantage.
Roger Williams
Senior Business Consultant
Decision Analytics
Experian
This article is the first in a series of optimisation related features that will appear in e-news.
| Related links |
| Read more client stories |
| Related links |
Contact us for further discussions about this article