Call us at 650-400-3029 (PST)

What the Future Holds for Decision Optimization

by | May 14, 2020 | AI, Decision Automation, Decision Management, Machine Learning, Strategy

A Guest Post by Neill Crossley, ACIB

James, thank you for the opportunity to guest blog in your series on Decision Optimization. First to introduce myself…..

I’m a veteran of 35 years in Retail Financial Services, 25 of which working in Decision Management, the last 17 focused on Decision Optimization, I currently work as a Product Manager in FICO’s Optimization group, focused on the software solutions we develop for business users & business analysts.

As James has discussed, optimizing decisions can be complex, requiring the management of many conflicting trade-offs, but often with huge benefits. Today’s analysis, modelling, and optimization tools can provide great insights into what influences your decision outcomes. In turn, these can help make better decisions that meet your objectives, as well as anticipate changes in the market or economy. Even so, many of these tools are still in their relative infancy.

Part of my role is to look to the future, at what users will expect and what technology can deliver. Recent studies identified 5 key themes, summarizing future needs:

  • Ease of Use – software should be intuitive, simple to set up, use and visualize key information
  • Automation – laborious tasks happen automatically, key tasks are guided by technology
  • Accessibility – easy sharing of data, models, and assets
  • Built-in Lifecycle Support – monitoring results & alerts, refreshing models, governance & approval
  • Adopt Innovation – take advantage of the latest and greatest technologies

In summary, business decision makers want to be quickly and accurately informed about solutions to complex problems, but also be agile to learn and continuously improve. Systems should be designed to be automated, but also easily augmented and controlled by business domain experts. But what might this look like?

Below are some examples of the potential user experience in the next generation of decision optimization software, based on a Business Analyst reviewing new Collections decision strategies.

Setting Targets, Goals & Constraints

  • After reviewing a range of Key Performance Indicators (KPIs), the analyst selects one or more primary targets, and sets the desired values.
  • They also select and define secondary goals, they need or want to meet, for example, the Number of Collectors available for use.
  • In effect, the analyst is defining parameters for a series of optimization scenarios, formed from combinations of the selected targets and goals. To identify the best decisions for each customer, a mathematical optimization model is employed to optimize the multiple objectives for each potential scenario.

Understanding Potential Solutions

  • The results for each scenario are displayed in various ways, for example, as Efficient Frontiers, plotting each scenario as a point on a chart, whose axes are KPIs such as Profit & Loss.
  • The scenarios are grouped in families identifying they are connected, and which goal or constraint is being incremented.
  • The user can click on any point on the chart, to review and compare each scenario’s resulting KPI values, against business as usual, and across different populations.

Decision Strategy Review

  • Once a potential new scenario has been identified, the analyst can also review options for operationalizing it as a decision tree or rule set.
  • For each scenario, a range of different decision strategies are automatically created, using techniques such as the global tree optimization approach James discussed in his last blog.
  • These strategies can be viewed, analyzed, edited and further refined, before being deployed, with one click into a production decision management environment, as a challenger decision strategy.

Monitor & Review

  • Over time, the tool automatically tracks the challenger strategy, to confirm it is performing as expected and compare it to other strategies.
  • In turn, new and additional data from the different strategies is captured. Which feeds into the automatic refresh of the underlying predictive, prescriptive (AI/ML), decision and optimization models. These feed into the next iteration of the Tool, which with improved accuracy, helps to identify further opportunities, continuing to iterate on an ongoing basis.

I hope these visualizations help you to consider the potential value of Decision Optimization tools, as they continue to evolve to meet user needs and take advantage of the latest technologies.

Feel free to contact me at, referencing this blog, should you want to understand more about “What the future holds for Decision Optimization”. Thank you.

Don’t forget you can also download the new paper and review the other posts in this series:


Join our Newsletter