Conduct UX Research and Test Early Concepts
This domain tests planning and running usability studies and turning raw observations into insights that stakeholders can act on - without contaminating the data through leading questions or bias.
Planning the study
A research plan states project background, research goals, research questions, KPIs, methodology, participant criteria, a moderator script, and a timeline. Choose moderated studies for rich follow-up questions and unmoderated studies for scale and natural context. Qualitative methods answer why; quantitative methods answer how many.
Running a usability study
Five participants per round typically surface most usability problems; recruit people who match the target user, obtain informed consent, and record only with permission. Moderators stay neutral: ask open questions, avoid leading wording, never teach the interface mid-task. Track task success rate, time on task, error rate, and satisfaction (for example SUS).
Synthesis and reporting
Collect observations, group them in an affinity diagram, name the themes, and write insights. An insight is an interpretation supported by observations - not the observation itself, and not yet the recommendation. Report by severity and frequency, tie findings to the research questions and KPIs, and share with stakeholders.
Bias in research
Observer and confirmation bias filter what you notice. Social desirability bias makes participants say what pleases you. The Hawthorne effect means people behave differently when watched. Leading questions (“Don’t you think the button is hard to find?”) manufacture the result you get.
Worked example
A checkout study asks where users abandon. Five moderated remote sessions; tasks: add item, apply promo code, check out. Observations: four of five miss the promo field; three hesitate at shipping options. Grouping in an affinity diagram produces the themes “promo discoverability” and “shipping clarity.” Insight: “Users do not notice the promo field because it is visually subordinate to the order summary.” Recommendation: move and restyle it, then re-test. Success rises from 60% to 90%.
Traps
Leading questions. Reporting opinions as observations. Changing the prototype mid-study so participants see different designs. Skipping consent or exposing participant data. Concluding from one participant. Writing a recommendation and calling it an insight.