Turning operational data into decision visibility
Turned fragmented operational data into a decision-support experience that helped teams understand performance, bottlenecks and where attention was required.
The problem
Operational data already existed, but users lacked a consistent way to understand overall performance, bottlenecks, incomplete or outstanding work and where attention was required.
The core problem was not lack of data. It was lack of usable decision visibility.
I led the product work across problem definition, stakeholder discovery, decision framing, metric definition, information hierarchy, requirements, validation and iteration.
- Different stakeholders needed different levels of information.
- Available data was not automatically decision-useful.
- Metrics needed consistent definitions across views.
Aggregate views could hide operational problems, while too much detail could create information overload. The product needed to help users move from overview to investigation without forcing them to interpret every available data point at once.
The product decision
The key product shift was from asking: “What data can we show?” to: “What decision does the user need to make?”
Data availability ≠ decision visibility.
Decision first
Define the decision first, then the metric.
Signal before diagnosis
Separate overview signals from diagnostic detail.
Consistent definitions
Use consistent metric definitions across views.
Attention management
Design for attention management, not maximum information density.
- Operational data
- Defined metrics
- Meaningful signals
- Prioritised views
- User decision
How I approached it
Understand
Identified the recurring questions stakeholders needed the product to answer.
Define
Translated those questions into measurable product concepts and consistent metric definitions.
Structure + Validate
Designed the information hierarchy and validated proposed metrics against underlying product data.
Deliver + Iterate
Worked with technical and operational stakeholders through implementation, clarification, launch and iteration.
What I learned
More data does not automatically create a better analytics product. A useful data product starts with the decision the user must make, then defines the signal required to support it.
Metric definition and information hierarchy were product decisions, not merely reporting choices. In this work, the value came from shaping how users moved from signal to diagnosis, not from exposing more data for its own sake.