user researcher

After a public breach, self-reported trust in the security center cratered to 32%. The team needed to know whether more reassurance copy would help — or if trust required something else entirely.

Fourteen customers, three weeks — contextual interviews paired with a short diary study tracking how people actually engaged with security notifications day to day.

The finding that mattered: people didn't want more security theater, they wanted to see exactly what had changed. A visible, itemized change log took trust from 32% to 78%.

Landing at 100%, account setup at 74%, first project at just 38% — half of signups were vanishing before they ever created a project.

Twenty-two moderated usability sessions traced the drop to one screen: step three asked people to name a project they hadn't made yet — a small ask with an outsized cost.

Moving project naming after the first action instead of before it took activation from 38% to 61% — nearly doubling it without touching the design system.

A scheduling app used across English and Spanish clinics had a translated interface — but no one had checked whether the underlying mental models translated too.

Thirty-one participants sorted the same set of terms twice, once per language, surfacing where categories held together and where they quietly fell apart.

The medical terms translated fine; the trust cues didn't carry over at all — reshaping how the team labeled anything touching cost and consent, in both languages.