Without a defined decision, discovery can become a collection of interesting observations that leave the roadmap untouched.
I agree the question with engineering, design and commercial colleagues before the work begins. The result should clarify what to build, what to test next or what not to pursue. Evidence is useful when it changes the organisation's commitment.
Discovery drifts for reasons that are easy to sympathise with. Researchers are curious, and customer conversations always surface something interesting. Teams under pressure to look busy can present a stream of insights as progress. Meanwhile, the roadmap continues on its own momentum, because nobody has connected the research to a point where a commitment would be made or withdrawn. The organisation ends up paying for two activities, discovery and delivery, that run in parallel without informing each other, and both teams grow frustrated with the other.
One practice I find helpful is to size discovery against the investment it informs. A small decision deserves a short, light exercise; a large or irreversible commitment justifies more rigour and more time. I also ask the team to choose the cheapest method that could change our view, whether a handful of interviews, a prototype, an analysis of existing usage or a simple test of demand. Matching effort to the size of the decision keeps discovery proportionate and makes its cost visible alongside the cost of building.
Consider a team debating whether to build a reporting module that several customers have mentioned. Rather than starting with broad interviews, the team agrees that the real question is whether customers would change their workflow to use reports inside the product rather than exporting data elsewhere. A prototype shown to a small group of users, together with an analysis of current export behaviour, may answer that quickly. If the evidence is weak, the team can redirect its effort before significant engineering work has begun, and the deferral is easy to explain.
The objection I hear is that tying discovery to decisions will narrow it, causing teams to miss the unexpected insight that open exploration provides. I think the concern is valid and I leave room for it. Some research time can remain exploratory, provided it is labelled as such and reviewed for patterns that might justify a future decision. What I resist is exploratory work presented as if it were evidence for a commitment. Keeping the two distinct protects both the curiosity and the discipline, and it keeps each honest about its purpose.
This asks commercial and engineering leaders to take part in discovery rather than receive its conclusions. When they help frame the question and observe the evidence, they are more likely to accept an inconvenient answer. It also asks executives to tolerate decisions not to build, which can feel like lost momentum. For the board, the useful signal is not the volume of research activity but how often significant commitments were shaped, redirected or stopped by evidence before substantial engineering effort was spent.
I ask every discovery effort to state, before it starts, what decision it will inform and what evidence would change our minds. That discipline keeps research focused and protects teams from endless exploration. It also makes the eventual decision easier to explain to the people who will build and sell the result.