Central Thesis
Most product teams are great at delivery and terrible at discovery. Torres's argument: the gap between what companies build and what customers actually want exists because discovery is treated as a one-time event, a research phase at project kickoff, rather than a continuous habit. The fix is weekly customer touchpoints, not big quarterly studies, run by the team building the product itself, in pursuit of a specific desired outcome.
"At a minimum, weekly touchpoints with customers, by the team building the product, where they conduct small research activities, in pursuit of a desired outcome."
Who Does Discovery: The Product Trio
Discovery belongs to a product trio, product manager, designer, and software engineer, not a dedicated researcher or UX team. The PM brings business context, the designer brings user empathy, the engineer brings technical feasibility. The trio can expand, but decision speed drops as headcount rises, so the group stays deliberately small. Problem space and solution space co-evolve; severing design from engineering breaks that feedback loop.
6 Prerequisite Mindsets
Outcome-oriented (success is impact, not features shipped), customer-centric (customer needs sit alongside business needs, not beneath them), collaborative (cross-functional decisions, not hand-offs), visual (draw and map rather than just discuss), experimental (identify assumptions, gather evidence like a scientist), and continuous (discovery as a permanent cadence, never a project phase).
The Opportunity Solution Tree
The OST is a visual map of the paths a team might take toward a desired outcome, with four levels: the desired outcome (a business need), the opportunity space (customer needs, pain points, desires), the solution space (ideas being explored), and assumption tests (how to evaluate solutions quickly).
The tree resolves the tension between business and customer needs, builds shared understanding across the trio, breaks large opportunities into smaller iterative slices, and replaces "should we build this?" with "which of these options is best?" Wells Fargo is the cautionary tale: it picked the right business outcome (accounts per customer) without a customer-centric filter, and bankers gamed the metric instead of serving customers. An outcome without an opportunity-space filter is dangerous.
Habit 1: Outcomes Over Outputs
| Metric type | What it measures | Best for |
|---|---|---|
| Business outcome | Financial or strategic health | Executive context-setting |
| Product outcome | How well the product moves the business forward | Product trios, real autonomy |
| Traction metric | Usage of a specific feature | Optimization, not open discovery |
Assign product outcomes to trios, not business outcomes (too broad) or traction metrics (too narrow). Lagging indicators like 90-day retention are hard to act on directly, find leading indicators that predict them. When facing a genuinely new outcome, start with a learning goal ("discover what drives churn") before committing to a performance goal ("reduce churn 10%"). Work one outcome at a time, ping-ponging quarterly means never climbing the learning curve.
Habit 3: Continuous Interviewing
The purpose of interviews is not asking customers what to build, it's discovering opportunities. Direct questions fail because people are bad at reporting their own behavior: they overestimate how often they do things they value, and describe their ideal self rather than their actual self. Torres's own product built for what recruiters said they wanted (sourcing passive candidates) flopped, because recruiters actually sourced active candidates, since that's what they were measured on.
The fix is story-based interviewing: ask for a specific, recent story rather than an opinion or hypothetical. "What criteria do you use when buying jeans?" invokes an idealized self; "tell me about the last time you bought jeans" reveals actual behavior. Interview at least one customer a week as a permanent habit, capture a one-page interview snapshot right after each session, and synthesize continuously rather than waiting to batch twelve interviews together.
Habit 4: Mapping the Opportunity Space
A flat backlog can't meaningfully compare opportunities of different shapes and sizes. Structuring them as a tree reveals parent-child relationships (one opportunity as a subset of another) and sibling relationships (distinct opportunities at the same level), which decomposes intractable problems and enables genuine compare-and-contrast prioritization instead of one-off yes/no decisions.
| Anti-pattern | Fix |
|---|---|
| Company-perspective framing ("I wish I had more subscriptions") | Reframe from the customer's point of view |
| Solutions in disguise ("I wish I could fast-forward ads") | Ask if there's more than one way to solve it; if not, reframe to the underlying need |
| Feelings as opportunities ("I'm frustrated") | Capture the actual cause behind the feeling |
Habit 6: Supercharged Ideation
Quantity predicts quality, the most original ideas usually surface later in a session, not first. Traditional group brainstorming underperforms individual ideation due to social loafing, group conformity, production blocking, and downward norm-setting, even though it feels productive (the "illusion of group productivity"). The fix is brainwriting: individual ideation, group sharing without evaluation, individual ideation again using others' ideas as inspiration, repeated until 15–20 ideas exist, then dot-voted down to three, not one, viable candidates for genuine comparison.
Habits 7–8: Assumptions and Testing Them
Every idea rests on assumptions across five types: desirability (will customers want this), viability (should we build it, does it work economically), feasibility (can we build it), usability (can customers actually use it), and ethical (could it cause harm we haven't considered). Portland once built condos with 1–2 bedrooms for displaced families who mostly had four or more people, an untested assumption that wasted tens of millions of dollars.
The fix is testing assumptions, not full ideas: fake-door buttons, landing pages, paper prototypes walked through in fifteen-minute interviews, one assumption per test. The best teams run 10–20 assumption tests a week (Marty Cagan), which is only possible because they're testing fragments, not shipped products.
Habit 10: Managing the Cycles
Discovery isn't linear, teams move up and down the tree as they learn. A shallow opportunity space means more interviews are needed; a sprawling one means narrowing focus; too few solutions means running an ideation session; being stuck on an opportunity that isn't moving the metric means questioning whether it was the right opportunity at all. The best teams work both top-down (outcome → opportunity → solution → test) and bottom-up (test results reshape the opportunity framing).
Key Mental Models
The single most important decision-making shift in the book is moving from "whether or not" ("should we build this feature?") to "compare and contrast" ("which of these three options best addresses this need?"). An opportunity, by definition, has multiple possible solutions, if it only has one, it's a solution wearing an opportunity's clothing.
"The economic value of any company is a function of its ability to continuously discover products that create value for customers in a way that creates value for the business."
Quick-Use Summary
The idea in one sentence: discovery should be a weekly habit run by the people building the product, not a research phase that happens once before delivery starts.
The three most applicable concepts:
- The Opportunity Solution Tree, structuring the messy space between a business outcome and a shipped feature.
- Story-based interviewing, asking for a specific recent story instead of an opinion, since people are unreliable narrators of their own behavior.
- Testing assumptions before building, the cheapest possible way to catch a wrong idea before engineering time is spent on it.