The Opportunity Solution Tree

The Opportunity Solution Tree (OST) is Teresa Torres' answer to the most common discovery failure I see: a team that does customer research, collects a pile of insights and feature ideas, and then has no structured way to decide which to pursue, so they default to whoever argued loudest or shipped fastest. The OST is a visual map that forces every solution to earn its place by tracing back to a customer opportunity, and every opportunity back to a business outcome.

#The four levels

It's a tree read top-down, and the structure is the argument:

            OUTCOME            (the one business result we're driving)
               │
        ┌──────┼──────┐
   OPPORTUNITY  OPPORTUNITY  OPPORTUNITY   (customer needs / pains / desires)
        │
   ┌────┼────┐
 SOLUTION SOLUTION SOLUTION   (ideas that address an opportunity)
        │
   ┌────┴────┐
  TEST     TEST              (assumption tests that de-risk a solution)

1. Outcome (root) is a single business or product outcome you are trying to move, for example "increase week-1 retention". Not a feature, not an output, but an outcome. The tree has one root on purpose.
2. Opportunities (branches) are customer needs, pain points, and desires surfaced through research, framed as opportunities to improve the outcome. Crucially these are framed in the customer's terms, not as solutions. "Users can't tell which bets are live" is an opportunity. "Add a filter" is not.
3. Solutions (sub-branches) are ideas that might address a specific opportunity. By hanging each solution off an opportunity, the tree quietly kills the orphan feature that serves no real need.
4. Assumption tests (leaves) are the small experiments that de-risk a solution before you commit to building it. This is where the tree connects to Experimentation and A-B Testing and rapid prototyping.

#Why I find it powerful

Three things it does well:

• It keeps discovery anchored. This is the companion to Continuous Discovery, the "in pursuit of a desired outcome" clause of Torres' definition made visible. Weekly touchpoints fill in the opportunity space, and the tree keeps them from wandering.
• It separates the problem space from the solution space. Opportunities are problems and solutions are answers. Most teams collapse the two and jump straight to features. The tree forces you to map the problem space first and pick the biggest opportunity before brainstorming solutions.
• It compares opportunities, not features. This is Torres' key move: prioritize at the opportunity level. Comparing "ship feature A vs feature B" is apples to oranges. Comparing "which customer problem matters most for this outcome" is a decision you can actually reason about.

#How it differs from a roadmap or a backlog

A roadmap is a time-ordered list of what you will ship, and a backlog is an undifferentiated pile of solutions. The OST is neither. It is a reasoning structure that shows why a solution connects to a need connects to a goal. You can generate a roadmap from a tree, but the tree shows the logic a roadmap hides. This pairs naturally with OST-driven discovery feeding roadmap decisions.

#Where I'm careful with it

I have watched OSTs become beautiful, sprawling diagrams that nobody updates and no decision ever flows from, discovery theater in tree form. The tree is a thinking tool, not a deliverable to be admired. Keep it small, keep it tied to one live outcome, and let it die when the outcome changes. A tidy tree that does not change what you build is worth nothing.

#How AI is changing it

AI can speed up the population of the tree: clustering interview transcripts and support tickets into candidate opportunities, and generating a spread of solution ideas for any branch far faster than a brainstorm. That is genuinely useful for the breadth. But the judgment the tree exists to support stays human. Which opportunity is highest-leverage, and whether a machine-generated cluster is a real need or just an artifact of how you phrased the query, is your call. AI fills the branches. You still decide which branch to climb. See AI in Discovery and Research.

#Continue Reading

• Continuous Discovery for the cadence the tree organizes; both are Teresa Torres' work.
• Outcomes Over Outputs for the outcome that sits at the root of every tree.
• Prioritization Frameworks for how to choose between opportunities once they are mapped.
• Experimentation and A-B Testing for the assumption tests that form the leaves.
• Jobs To Be Done for another lens on framing the opportunity space.