The tool had structure
What it needed was confidence

Four weeks, a wireframe, and a lot of assumptions

The engagement was four weeks

The brief was to take a research wireframe and turn it into a clear, compliant, build-ready prototype for a due diligence tool used across multiple government departments

There was no access to the original tool. A wireframe showed the intended direction, and screenshots provided some context. That was everything to work from

Which meant a lot of the early work was about testing assumptions: inherited ones, made ones, and ones nobody had thought to question yet

Four weeks to rebuild a tool I'd not seen

The assumption the wireframe had made

The wireframe had identified real structural problems: nested content, unclear hierarchy, no consistent flow. The proposed solution added organisation and depth

But organising something that lacks clarity doesn't always produce confidence. And confidence was what the users actually needed

That took until week three to become fully visible

Structure and clarity aren't the same thing

What users said in testing

Daily iteration had been shaping the prototype from the start

By week three, user testing sessions brought something more direct: real users, real workflows, real friction

One user said it plainly:

But the deeper pattern across sessions was this: users were being asked to trust data, recommendations and outcomes, without being told why those things were trustworthy or what they meant for next steps

The tool was producing outputs
It wasn't explaining itself

The insight

Users didn't need more features

They needed more confidence

Not confidence in the sense of reassurance, confidence in the sense of clarity

Evidence behind every outcome. Context at every decision point. A system that showed its working so users could act on it without doubt

That reframed everything that followed

What got built

Unmatched records could be refined in place, surfacing results that would otherwise be missed

Confidence scoring made match certainty visible at a glance

Clear status flags told users exactly where to focus next, without anything outside the tool to interpret the outcome

Every state, every edge case, every unhappy path: tested and resolved before handoff

Four weeks
Build-ready

The payoff

Hours spent on manual workarounds, gone

And the broader feedback confirmed what the testing had suggested: when users understood why a result was trustworthy, they could act on it with confidence

That was the gap the wireframe hadn't named

Naming it was half the job
Closing it was the other half