Generative research: discovery before design

Generative research answers the questions that come before design: how do customers actually work, what are they trying to accomplish, what do they believe about how things function, what context surrounds the task? It is discovery, not judgment. It does not evaluate a design or validate a plan, because there is nothing to evaluate yet. Its output is the raw understanding that determines whether the thing you eventually build is the right thing at all. Expect one to three months of work, rarely less than one.

Why it matters to the business

Building on assumptions instead of discovery is how delivery gaps happen. Bain's survey of 362 firms found 80% believed they delivered a superior experience while only 8% of their customers agreed. That gap starts upstream, in what companies think they know about customer behavior but never actually observed.

Asking is not enough, because people misreport their own behavior. An 84.51 study comparing survey answers with loyalty-card data found 75% of respondents misstated their actual purchases. Generative methods that watch behavior, such as observation and diary studies, close that say-do gap before it becomes a product bet.

How to use it

• Schedule discovery one to three months ahead of the design cycle so it feeds the roadmap instead of blocking a sprint.
• Run observational studies: watch what people do, not what they say they do.
• Use in-depth interviews to probe mental models and habits, not to pitch concepts.
• Deploy diary studies for longitudinal behavior; they surface forgotten steps and workarounds no single interview catches.
• Keep validation agendas out of the room; the moment a design needs approving, it is a different study.

Where teams get it wrong

Teams run generative research as validation in disguise: a design already exists, and the sessions exist to bless it. Or they compress discovery into a two-week sprint, run three thin interviews, and declare the assumptions validated. Both produce the appearance of customer understanding with none of the substance, which is worse than knowing you have not looked yet.

Ask your team

• What did we learn about customer behavior before we designed this, and by what method?
• Which of our findings come from watching behavior rather than asking about it?
• Is next quarter's discovery work scheduled ahead of the design cycle, or will it be squeezed into a sprint?

The purpose is discovery, not judgment.