Building a Feedback Loop from Operational Data

1. Context and Challenge: Making the System Concrete

We needed a clear way to explain how operational data could improve our policies and decisions. Financial results showed what the business had achieved, but they did not show how the underlying systems worked or how they could improve over time.

The demonstration therefore had to distinguish among three things:

1. What works today: Components already collecting or using operational data.
2. What remains in development: Connections and tools that are not yet complete.
3. How the parts fit together: A planned feedback loop from daily work to better decisions.

---

2. Action Taken: Building the Demonstration

We combined working software with a planned architecture. The working components made the idea tangible, while the design showed how unfinished pieces would connect.

We organized the demonstration around three steps:

1. Map the work: Identify important process steps and decisions.
2. Record useful data: Capture structured data and context at those points.
3. Review and improve decisions: Use the records to test policies, reduce variation, and shorten processing time.

graph TD
    P1[Daily work] -->|Record data| P2[Structured records]
    P2 -->|Review outcomes| P3[Models and policy review]
    P3 -->|Update decisions| P4[Improved rules and tools]
    P4 -->|Guide future work| P1

This flow gave the audience a concrete way to see how daily operations could produce data that improves future decisions.

---

3. Result and Next Steps

• Completed Demonstration: Built and presented a demonstration that combined functioning components with the planned system design.
• Clearer Explanation: Used the working example to make the proposed feedback loop easier to understand, without treating audience understanding as a measured outcome.
• Implementation Status: Confirmed which components already work and which connections still need to be built.