CX Reference Knowledge · Research & Discovery
CX Research Methods & Quality Standards
VOC tells you what is happening, rarely why. Research designed to prove an idea is not research.
Research quality is not about doing research; it is about doing it right. Quality depends on planning, method fit, recruitment, neutral moderation, and honest analysis. Generative research finds the problem, evaluative research tests the answer, and the order of work matters.
Ten things decide whether research is worth acting on: planning that challenges assumptions, methods chosen for the question rather than the budget, the right participants in sufficient numbers, neutral moderation, questions that neither lead nor ask people to predict, real time invested in synthesis, actionable output, an actual answer to the original question, separation of flukes from signals, and no self-fulfilling design. One quote is not an insight.
- Generative: observation, interviews, diary studies; 8-12 per segment; 1-3 months
- Evaluative: tree testing, usability testing; minimum 5 per segment, fresh each round
- Never test experiences users cannot realistically perform
- Include an accessibility group in recruitment
VOC — surveys, reviews, tickets, ratings — tells you what is happening, rarely why. It needs augmentation with observational research, interviews, task analysis, and contextual inquiry. The correct order of work: generative research, problem definition, strategy, prioritization, concepts and prototypes, testing and iteration, delivery, then post-release monitoring. Skipping the front of that chain is how teams build the wrong thing well.
Apply this
Reading about cx research methods & quality standards is one thing. Seeing where it applies in your journey is the useful part.