I start by getting close to the problem — talking to stakeholders, reviewing existing data, and identifying what we actually need to learn versus what we think we already know. From there I build a research plan that balances speed with rigor, choosing methods based on the question rather than habit.

Ambiguity is usually a signal that the right problem has not been framed yet. I treat early-stage work as hypothesis generation — running quick exploratory research to surface assumptions, map the system, and find the lever points worth investigating further before committing to a solution direction.

It depends on the question. For understanding mental models and decision-making, I lean on semi-structured interviews and think-alouds. For evaluating design decisions, usability testing and heuristic analysis. For broader patterns, I combine surveys with behavioral data. I am mixed-methods by default — qual tells you the why, quant tells you the how much.

Research only has impact if the people building the product believe in it. I involve designers, PMs, and engineers early — in research planning, as observers during sessions, and in synthesis. When the team has seen the user themselves, the findings land differently.

A report nobody reads is not good research. Good output is whatever format makes the insight impossible to ignore — a one-pager, a decision framework, a short video clip of a user session. The goal is action, not documentation.