Product Discovery

Updated 2026-04-09

The process product teams use to figure out what should be built before they build it. Unlike classic requirements engineering, discovery is iterative, customer-driven, and never really ends. It is not a project phase. It is an ongoing habit.

The Problem with One-Off Research

The usual pattern looks like this: before a launch there is a discovery sprint with interviews, personas, and workshops. Then the team builds. Then the discovery budget is gone. The result is a product optimized for a user model from six months ago. Teresa Torres calls this “batch discovery,” and it is one of the main reasons product teams build the wrong thing well.

Continuous Discovery

The Torres baseline is simple: at least one customer interview per week, per team. Not as a special research project, but as routine. That creates:

  1. A current and real picture of users instead of stale personas.
  2. Faster calibration of hypotheses.
  3. Better intuition for nuance, which only forms through repeated contact.

The central tool is the Opportunity Solution Tree (OST), a visual structure that connects outcome → opportunities → solutions → experiments. It prevents teams from jumping directly from user feedback to implementation ideas.

Discovery Under AI Pressure

With AI tools such as Cursor and GitHub Copilot, the question shifts: if a PM can prototype or build more directly, do we still need the classic PM-design-engineering trio during discovery? Teresa Torres and Petra Wille argue that the ability to build faster does not replace understanding what should be built. Discovery remains human because nuance in customer conversations is still hard to extract reliably with an LLM.

Their updated answer to the role question is Product Builder: fewer rigid handoffs between PM, design, and engineering, more shared foundational understanding. That makes discovery more important, not less. The faster you can build, the faster you can build the wrong thing.

Connections

  • Teresa Torres — central figure in modern discovery thinking, and the key starting point.
  • Petra Wille — coaches discovery practices in product leadership contexts.
  • Product Builder — describes the new role logic being renegotiated under AI pressure.
  • Lenny Rachitsky — regularly publishes empirical data on discovery practices in tech teams.
  • AI Evals — discovery is also the right mindset for evaluation design: both start from real user problems instead of assumptions.

Sources