Better Business Decisions, One Cup at a Time
Researchers at Cornell University estimate we make 226.7 decisions each day on food alone. Crazy, right? But if we attempted to model the decision of what coffee type to order just for fun, perhaps we could validate the results and advance science?
In building the model, you would likely consider basic personal preferences for flavor, texture, temperature and strength of the beverage and perhaps augment it with a variety of other factors dealing with social considerations, like fair trade, and physical attributes like reaction to caffeine. The ultimate list of considerations, or potential decisions, could become unwieldy.
Enter decision management. DM is a relatively new category of software that applies a decision model to logic, such as all the factors people consider before ordering a cup of coffee. Think of a decision tree followed in reverse where each layer of factors ultimately narrows to a single outcome – say, giving a Starbucks barista an order for an iced grande latte with soy milk and caramel drizzled in a perfect grid pattern.
The larger point is that these outcomes are based on the number of choices or decisions available to us, which gets complicated, especially in large organizations. Managing these decisions is the domain of decision management, and it’s used everywhere, from the largest banks, insurers and retailers to healthcare organizations and governmental agencies.
Sapiens recently introduced a significant new release of Sapiens Decision, our no-code decision management platform, with a launch event that featured a modeling challenge for client teams to demonstrate how they would model a specific decision. The cross-client teams collaborated on their challenges via our cloud-provided platform and had the opportunity to test out the new user experience, notation, and enhanced functionality. The “Decisive Enterprisers” team, anchored by Sapiens Decision users from Allstate, Wells Fargo and Freddie Mac, took on the challenge of modeling our coffee order quandary described above, going toe-to-toe for bragging rights against other cross-client teams that brought similarly inspired use-cases.
The results were creative, insightful and, quite frankly, a little scary. In just a few hours, the team constructed a complete decision model to determine specific facts and apply them through rules to generate the coffee order.
Screenshots of the model are shown below, and in dissecting the model, the team split the decision rules into two main branches: order size and order type. The next layers below describe exactly what factors define order size and type. This is where the team’s laser insights become a bit unsettling.
Here’s a sampling of the facts to be determined within the model:
- White shirt indicator – spill prone?
- Caffeine sensitivity indicator
- Prior coffee consumption count in cups
- Work time remaining in hours
- Outdoor temperature in degrees
- Season code
- Weather condition code
- Wind speed
- Drink recipient identifier – self, friend, boss
- Spend occasion indicator
- Food sensitivity code
- Diet program code
- Home bathroom commute
- Paruresis indicator – i.e. “shy bladder”
The graphic overview of the decision model is about as technically complex as it gets for the modeler and their stakeholders, with supporting rule family detail available behind the notation that is similarly intuitive whether you’re modeling coffee orders, an insurance claim process, or mortgage approval.
And, unlike traditional business rules tools which are technical, require developers to interpret written requirements, and quickly generate a mess of legacy code that is impossible to manage, Sapiens Decision provides an easy-to-understand and easy-to-manage repository that remains stable over time.
Business analysts who model rules using spreadsheets transition seamlessly to our platform. That’s why we refer to our platform as “no code”. Business analysts don’t need to be coders, just logical thinkers. Build the model, use the built-in validation tools to avoid gaps and conflicts, test robustly, change as needed, approve, then deploy auto-generated (and error free) code. IT is freed up to solve other challenging engineering problems.
And unlike our modeling challenge, business analysts don’t always need to build decision models from scratch. Organizations already have applications with millions of lines of code containing decision logic that requires constant updating. We provide automated tools to extract logic from existing applications with the ability to process code written in everything from COBOL to Drools.
Getting back to our original question, our Decisive Enterprisers team broke down the coffee order process into decisions encompassing 17 facts. This translates to about 6% of our daily food decisions allocated to ordering a single cup of coffee, a reasonable result. Next time, let your barista know that fun fact and maybe you’ll get your caramel drizzled in a perfect grid. But maybe not.
Learn more about our Decision Management Solutions.