Validating the Trust Thesis at the Interaction Layer
| Property | Value |
|---|---|
| 📅 Project | OmniRisk · Financial Workflow Dashboard |
| 🔖 Phase | Test & Iterate |
| 🛠️ Method | Moderated remote sessions · Think-aloud · Post-task survey |
| ⏱️ Duration | 30–45 minutes per session |
| 👥 Participants | 3 — hedge-fund middle-office analyst, fintech middle-office analyst, FX trader |
This test was designed as a focused validation pass on the three trust decisions. The questions it set out to answer:
One — does the sequential workflow read as orchestration or as constraint? Sequential CTAs were the most opinionated workflow decision. Testing needed to confirm whether participants experienced the sequence as the workflow's natural logic or as a designer imposing structure.
Two — does the AI Recommendation modal earn trust at the moment of decision? The trust thesis predicted that confidence scoring and visible reasoning would change how operators engage with AI recommendations. Testing needed to surface whether the modal delivered on that prediction.
Three — does Manual Override read as a first-class action? Decision 03 paired Accept AI Recommendation and Manual Override as equal-weight choices. Testing needed to confirm whether the pairing communicated respect for operator judgment or read as a warning to be talked out of.
Four — does in-table highlighting communicate "AI recommendation" without explanation? In-table highlighting placed AI output inside the data that produced it as a Core feature. Testing needed to confirm whether highlighting alone was sufficient signal.
Five — does the workflow's phase progression communicate state to the user? Disabled-state CTAs were the navigation signal. Testing needed to surface whether disabled states alone communicated workflow position or whether additional affordance was needed.
The methodology was deliberately structured as a focused validation rather than open-ended usability research. The reasoning:
Research from Discover produced four convergent findings across four participants. Synthesis from Define translated those findings into design principles, personas, and a journey map. Design produced three trust decisions traceable to research evidence and the design principles. The remaining question was not "is the concept viable" — it was "do the interactions deliver on the concept's promise."
That question is best answered by a focused validation pass — fewer sessions, targeted protocol, specific hypotheses being tested. A discovery-oriented study would have been the wrong methodology when the design was already research-grounded.
| Role | Organization | Why Included |
|---|---|---|
| Middle Office Analyst | Hedge Fund | Primary user — tests whether the design works for the experienced operator |
| Middle Office Analyst | Fintech Startup | Scale comparison — tests whether the design assumes institutional infrastructure |
| FX Trader | Hedge Fund | Adjacent stakeholder — tests dashboard architecture for the front office view |
The head of IT (custodian bank) shaped research but was not part of testing. The prototype serves the people doing fund administration; the head of IT's systems vantage informed what the design had to support but is not the lens against which interaction quality is tested.