Improving the quality of your decision-making can seem straightforward. Take fraud. You know what a good decision looks like – one that results in less fraud. Working to change your decision in ways that reduce the amount of fraud should be easy. Each decision strategy can be assessed to see how much fraud results and the one that results in the least fraud can be adopted.
If only it were so simple. We once had a client that thought this way. They engaged us to help them reduce fraud in their credit card business. In fact, they kicked off the project by saying that reducing fraud was the only metric that mattered to them – we just had to reduce the total value of fraudulent transactions. We looked at each other, smiled and said
“We can reduce credit card fraud to zero”
They were super excited, of course, and asked how?
“Don’t accept credit cards”
After a short pause they said that, while fraud reduction was the goal, obviously this could not come at the expense of their entire business. And with that we could have a more sensible conversation about their goals and the trade offs inherent in their business.
And this is a common problem. Having many trade offs is common in making decisions. A good decision is not just one that reduces fraud, it must also avoid upsetting good customers, not result in too many calls to complain, doesn’t increase attrition, keeps the business able to operate, etc. etc.
All these trade offs make it increasingly different to find the “best” approach to a decision. And even if, somehow, you manage to wrap your brain around the trade offs and come up with a “best” strategy, the world won’t stay still. Markets change, competitors change, fraudsters change their behavior, consumer expectations change and so what is “best” changes.
The solution is not to try and do this with just brain power. As we noted earlier, you can use models to define the relationships between the elements of your decision and your business results. You can run experiments to fill in your knowledge gaps. This creates a rich set of information with which to optimize your decision, but to exploit it you will need to use mathematical models to calculate the different permutations and an optimization engine or solver to find the best one.
Only this mathematical approach can manage all these trade offs and consider all the possible permutations to help you find the best results and the best approach. Once setup, you can realize huge business benefits, meet your goals and see multiple ways to improve your profitability.
But decision optimization is not a one-off exercise. Like all decision improvement it needs to be an ongoing process that considers alternative scenarios and new data while running experiments. This new paper outlines such a process. You can improve your decision-making in lots of ways. But to find an optimal approach, you’ll need to use math.
Other posts in this series:
- The Customer Journey to Decision Optimization
- Ongoing Decision Improvement Drives Decision Excellence
- Don’t Just Sit There, Experiment!
- Why Argue When You Can Model?