From a stalled MVP to an adopted decision product

The situation

An MVP already existed. It carried a great deal of data and a long list of use cases. The analysts and executives it had been built for were not using it.

What everyone thought the problem was

The brief could reasonably have been read as improve the dashboard. Better charts, clearer hierarchy, a tidier interface.

What I found

I interviewed analysts, operators and executives, and mapped the decisions they were actually making — supply and shortage monitoring, strategic reserve awareness, procurement calls, market assessment, executive oversight.

The product had been organised around the data that existed rather than around those decisions. It was not a presentation problem. People could not get from what the platform showed to what they had to decide, so they went back to the reports and spreadsheets they trusted.

The decision

Reframe the work around operational decisions rather than data coverage, and structure the experience by role instead of by data source. That meant questioning parts of the existing scope rather than restyling them.

What we made

Role-based workspaces for executive, reserve, trading and operations users. Market intelligence tied to the decisions it informed. AI-assisted querying shaped around the questions analysts actually asked, with clear provenance for the source data — trust being the precondition for any of it getting used.

What changed

The rebuilt platform went live. Analysts and executives who had abandoned the first version began using it in their operational routines, and it expanded to additional commodities. User research and decision mapping became a more permanent part of how the team worked.

What this proves

That a technically complete product people avoid is still a failed product — and that the fix is usually upstream of the interface.