investigate

Start with what you need to learn, then match the method. Use interviews or contextual inquiry for generative questions, usability tests or surveys for validation, diary studies for behavior over time, and card sorts or tree tests for information architecture questions.

Five to eight interviews typically reach thematic saturation for generative research. Five usability test participants per round catch roughly 85% of issues. Surveys need 100+ responses for descriptive statistics and 30+ per group for comparisons.

Ask about past behavior instead of hypotheticals, avoid leading or compound questions, and use open-ended prompts like "walk me through the last time you did X." Use silence and neutral probes such as "tell me more about that" rather than suggesting answers.

Moderated testing lets you probe confusion and observe behavior live, suiting complex tasks and early concepts. Unmoderated testing is faster and cheaper with larger samples, best for straightforward evaluative tasks on stable prototypes.

Use affinity mapping to cluster observations bottom-up, then apply thematic analysis to name patterns. Structure each insight as observation plus inference plus implication, and tag it with an evidence strength rating of strong, moderate, or weak.

Small qualitative samples reveal patterns but not prevalence. Findings from five to eight interviews are hypotheses about frequency, so validate with surveys or analytics before making claims about how common a behavior is across your user base.