Teresa Torres on Continuous Discovery
The core argument: Most teams treat customer research as a phase — a project that kicks off, generates insights, and then ends while the team moves on to building. Teresa Torres argues this is the wrong mental model entirely. Discovery isn't a stage in your process; it's a habit. Every week, the team should be talking to customers to inform the next decision, not a future roadmap cycle.
The Opportunity Solution Tree (OST): Torres's primary structural tool is the OST, which places your desired outcome at the top and branches downward through opportunities before ever reaching solutions. The critical discipline is decomposition: you keep breaking opportunities into smaller ones until you reach something you can actually address. The Netflix example she returns to — 'I can't find something to watch' → 'I don't know if this show's any good' → 'Who's in the cast?' — illustrates how evergreen, unsolvable problems become tractable through vertical decomposition. The tree also gives teams a strategic view of where they could play before committing to where they will play.
Cognitive limits on the tree: Torres invokes Miller's Law to argue that branching more than seven items at any level of the OST undermines its value as a thinking tool. At that point, the tree becomes a list — something you scroll past rather than reason with. The practical target is three to seven branches per level.
You don't need permission: One of Torres's most practically urgent points: individual contributors consistently overestimate how much organizational permission they need to talk to customers. If your company doesn't have a formal discovery program, that's not a blocker — it's an excuse. Torres's prescription is simple: carve out your own habit. Your company doesn't own you outside work hours, and you almost certainly know people who resemble your users. Start there.
Everything is already a bet: Torres reframes backlog skepticism directly. The intuition that items shouldn't enter the backlog without full discovery is false. Every backlog item is already a bet — discovery just makes the next bet better-informed than the last one. Crucially, Torres argues you should not halt delivery to wait for discovery to catch up. Run both in parallel. Stopping to 'properly discover' everything before shipping is itself a way to kill appetite for discovery inside an organization.