Imagine if you could directly connect the objectives and goals of the organisation to each and every customer decision you make… and get a performance improvement too. Whether that goal be profit, economic profit, shareholder value or simply portfolio bad debt, achieving it is made difficult by the number and complexity of decisions and constraints, and their ‘distance’ from the ultimate objective. This is a tough optimisation problem that Experian now has a simple solution for.

Over the past decade, many organisations developed excellent databases supporting predictive models that drove decisions deployed through efficient decisioning software. But even those that regularly refine and implement challenger strategy trees cannot be sure what opportunities are being missed, or how much that is costing in terms of lost profit and opportunity.

Optimisation in the mathematical sense describes making the correct combination of individual decisions to maximise an overall goal (such as profit) subject to the business constraints (such as accept rate or bad debt rate or operational capacity). The process involves trading off decisions so that the best group of decisions are made.

Traditionally, optimisation solutions within credit risk were difficult to implement, primarily because most decisions in the credit risk environment are made with the use of strategy trees, which doesn’t lend itself to making individual decisions. 

To tackle this challenge Experian developed a solution that can help a business user design an optimised strategy tree for seamless and rapid deployment into existing tree-based decisioning systems. This removes many implementation obstacles faced by organisations and opens up a wealth of opportunity to deploy optimisation across the different areas of credit risk, from Originations to Collection and Recoveries.

The solution enables a business analyst to formulate what the business wants to optimise (for example, profit) and specify the constraints that may occur at either a global, portfolio or customer level (for example, minimum time on book, maximum portfolio exposure) and then produce a new optimised strategy tree that makes the best set of decisions to achieve the business goal, in this case profit.

There are four key stages in the process,

• Define – one defines the key elements of the optimisation, for example, the business goal, KPIs and constraints.
• Optimise - one uses the software to optimise the strategy tree, creating new nodes if necessary, all facilitated by what-if scenario simulation. The optimisation engine uses proprietary mathematical techniques to rapidly find the optimal solution.
• Deploy - once the ‘best’ tree has been identified, it’s then deployed in the existing strategy tree-based decisioning software, so there is zero infrastructural impact. ‘Best’ is often a subjective decision, so may not always be the one that generates most profit, but one that is more palatable for an organisation initially. To aid this, users can configure multiple KPIs that help support the process of developing the ‘best’ tree.
• Maintain – finally, one just needs to monitor performance and ensure improvements are maintained.

Because the solution delivers new strategy trees, it means that there is no need to change any existing infrastructure. One can simply implement a newly optimised tree as a challenger. This means that the solution can be used alongside Strategy Management or, indeed, any other tree based decision tool.

So, optimisation now becomes a reality rather than just an aspiration and, with it, you can connect the strategic business goals to each and every customer decision. Whatever the goal, be it profit or bad debt for example, a double digit improvement in performance is a realistic expectation.

Adrian Carr
Senior Optimisation Consultant
Decision Analytics
Experian

Contact us for further discussions about strategy tree optimisation

Top