Around 80% of smart phone owners use navigation applications regularly. In fact, many of us are so used to using these navigation tools, we even use them when familiar with our planned destination.
For something so ingrained in our day to day lives, few of us have really stopped to think what these navigation apps are doing. A navigation app uses a combination of maps and GPS location data to provide turn-by-turn guidance from our current location to a destination of choice. Users set overarching objectives such as ‘fastest route’ or ‘shortest route’ and the navigation app will essentially consider all available routes to your destination to determine one that best meets your defined objective.
Let’s think on that for a second. The navigation system is assessing every possible route, considering the constraints of each leg of the route, such as traffic flow and legal speed limits, determining time and distance to destination for each route and then comparing all the considered routes against your set objective, to work out which one provides the best outcome.
Wouldn’t it be nice if we had something like this for navigating our business decisions?
We could tell it our corporate goals and ask it to help define a strategy that allowed us to meet those goals within our financial, regulatory or operational constraints.
In fact, we do. For many years FICO have been working with some of the world’s leading organisations to deploy similar techniques to those we use in our navigation systems to help solve significantly more complex business problems across a broad range of industries, including banking and finance, retail, manufacturing, telecommunications, auto finance, energy and supply chain management.
Take collections as an example. For each customer in collections, we have a range of options we could take each day, from do nothing, through to automated voicemails, SMS, emails or a more formal letter. Think of these as the possible routes we could take on a journey. Each of these treatment options has its own probability of success, which varies by customer and age of debt, along with its own cost. Just like different routes, each activity may cost more or take longer.
Evaluate All Your Options Upfront
In business, more progressive companies regularly look at challenger strategies to assess whether a better alternative is available and compare it with the current champion. While these decisions are often driven by deep domain experience, they are not much different to trying a different route to a favourite destination and seeing if it’s faster or slower. This can often take the business in the wrong direction and is a slow incremental process of improvement, towards an often-moving target.
By using powerful machine learning optimisation solvers in conjunction with existing statistical models and action-effect relationships, we can instead consider all the possible actions and likely effects, so that we have a full picture of possible outcomes and associated costs.
We can evaluate all the routes to our destination before we set off. Then, by overlaying business and operational constraints such as cost, affordability criteria, responsible collections practices and overarching business goals — such as maximise collections amount while minimising customer attrition or maintaining operation headcount —we can determine a collective set of actions across all customers that maximise our goals.
Managing Competing Business Objectives
Decision optimisation also enables different business units to input their own lower-level objectives and quantify the impacts and trade-offs. For example, one business unit may want to reduce operational costs, while another may want to increase focus by reducing the number of days until first contact.
Each objective can be introduced quickly into the system and the many millions of possible permutations of actions and outcomes reconsidered based on these new ‘journey’ constraints to find an updated optimised outcome. We can directly compare the trade-offs made to satisfy each condition, so that a consensus can be reached on the best overall solution for all parties.
This process often reminds me of family holidays with a number of ‘differing points of view’ on the best possible route to our destination, while also taking in points of interest along our way. The navigation app shows the impact of each new goal to see if it’s really worth the delay.
‘What If’ Analysis
Business users can also set soft constraints. Soft constraints typically include factors that we might be looking to flex or explore. An example might be that we have 50 FTE in the collections team and we want to establish the impact of reducing headcount or consider the diminishing returns of adding more headcount. This type of approach is invaluable in helping formulate business cases for financial or operational change.
As mentioned earlier, decision optimisation techniques have been used across a huge variety of business applications. I’ve listed just a few of those in financial services below:
- Loan Pricing
- Initial Credit Line
- Credit Line Increase/Decrease
- Early and Late-stage Collections
- Debt Recovery
- Next Best Action Marketing
- Deposit Pricing
In each case, decision optimisation has driven significant business improvement.
You wouldn’t take a journey to a new destination without using navigation. Your business decisions are even more critical.
Learn more about how to apply decision optimisation techniques from Petr Kapoun, CRO of Home Credit International, and the value it provided, in this video.
See more information on Petr's success with decision optimisation.