Continuous Discovery

Continuous discovery is the practice of maintaining regular, ongoing contact with customers throughout product development, rather than conducting research in discrete projects that precede delivery. Teresa Torres, who articulated the approach in detail, describes a weekly cadence of customer touchpoints conducted by the team building the product, with the resulting insight feeding decisions as they arise.

The argument against project-based research is practical. Research conducted as a phase produces a report whose findings are relevant when written and increasingly stale afterwards, and the understanding it generates resides largely with the researchers rather than with the team making daily decisions. Because a new study takes weeks to arrange, the many small decisions made between studies are made on assumption. A weekly rhythm changes the economics: a question that arises on Monday can be asked of a customer within days, so evidence becomes available at the speed decisions are made.

The insistence that the product trio, typically a product manager, a designer, and an engineer, conduct the interviews themselves is central rather than incidental. Insight loses most of its value in handover, since the nuance, the hesitations, and the surprising remarks that change a team's thinking rarely survive summarization. Teams that hear customers directly develop shared intuition, which improves the hundreds of small decisions that never reach a research question at all.

The approach is supported by structured artifacts, principally the opportunity solution tree, which connects a desired outcome to the customer opportunities that could produce it and to the solutions and experiments addressing each opportunity. Its function is to keep solution ideas connected to validated problems and to make visible the assumption chain behind a decision, which prevents the common pattern where a team accumulates feature ideas with no explicit link to any customer need or business result.

The obstacles are organizational rather than methodological. Recruiting participants every week is a logistics problem that defeats most teams unless it is automated through in-product prompts, customer lists, or standing panels. Teams under delivery pressure treat interviews as the first thing to cut. And in businesses where customers are few, senior, or reachable only through sales relationships, weekly access may be genuinely impossible, in which case a slower cadence with other evidence sources is the realistic adaptation.

The practice also changes how research findings are recorded, since a weekly cadence produces a continuous stream rather than discrete reports. Maintaining a searchable repository of interview snippets, observations, and the decisions they informed becomes necessary rather than optional, because without it the team accumulates impressions it cannot substantiate and repeats questions it has already answered. Tagging observations against the opportunities they relate to, and linking decisions back to the evidence behind them, also makes it possible to revisit a conclusion when circumstances change, which is difficult when the reasoning exists only in the memories of whoever attended the interviews.

Continuous discovery does not replace deeper research, and treating it as a substitute produces shallow understanding. Foundational studies, ethnographic work, and large-scale quantitative research answer questions that a weekly conversation cannot, and the two operate at different depths. In practice a mature user research function supports both, running the periodic deep work while enabling teams to maintain their own regular contact, and the resulting evidence feeds the prioritization and roadmap decisions handled through product research in a product studio setting.