Tested whether the ATM affordance vocabulary still made sense to people who had stopped using cash for daily purchases. Heavy cash users planned by bill count; light cash users planned by total amount. The existing UI privileged the bill-count model.
The redesign exposed both anchors simultaneously and recovered task success on denomination customization.
Notable null result: no significant difference between dark and light prototype conditions (46% vs 50%, p = .31). Reported explicitly rather than mining for a secondary win.
A redesign of the premier-tier mobile experience underperformed on the metric the business cared about most: time-to-advisor. The design team believed the metric was lagging. The business team believed the design was failing. Research had to settle that disagreement with evidence.
Reaction-time data confirmed the visual hierarchy was steering attention away from the action. Customers spent 14 seconds longer on the home screen before tapping anything.
Two product teams owned discrete phases of an account opening journey: discovery to sign-up, and verification to funding. Customers experienced the whole thing as disjointed, lacking a proper welcome as a new customer.
Compiled a journey from five moderated usability tests. It was quickly visible that the apply experience was disjointed from the onboarding experience, lacking a robust confirmation or celebration of task completion, relying on just a badly formatted email to bridge the experiences together. The solution I came up with was to facilitate a workshop between the two teams to address the gaps.
In the first months of COVID, hospital supervisors and regional executives needed to make decisions hourly with incomplete data. Existing dashboards were built for steady-state operations and broke down under crisis tempo.
Embedded with operations leadership for 14 months. Two-week iteration cycles against actual decisions being made that week.
Executives needed three numbers at a glance, in stable positions, with one clear signal for "is this normal." Supervisors needed the same data with a different temporal frame.
A B2B SaaS product had strong SEO landing performance but a steep drop at trial signup. The funnel team had hypotheses; nobody had talked to the users who dropped.
Combined Amplitude funnels with Hotjar session recordings to localize the drop, then ran 24 interviews split across users who completed signup, dropped at the form, and completed signup but never returned.
Watson OpenScale was building KPI and drift monitoring tools for model risk managers in financial services. The concepts had been informed by secondary research and SME conversations — but hadn't been validated with the model owners and validators who would actually live inside the workflow.
Four moderated remote concept walkthroughs with model risk managers, model owners, and validators across insurance, technology, and financial services. Participants narrated the KPI & drift monitoring flow against how they currently set up, review, and report on financial models — surfacing where the product's assumptions broke down.
Designed and ran design-thinking workshops across IBM, AIGA Design Week, AIGA Leadership Retreat, Aim High, and College Track. Audiences ranged from senior IBM designers to high-school students from underrepresented backgrounds.
Built a modular curriculum scaled from 90 minutes to two days. Always grounded in a real problem participants brought, never a synthetic exercise.
