Step 1
Define one desired outcome and map customer opportunities with explicit provenance.
Opportunity Solution Tree Builder
Map an outcome to evidence-backed customer opportunities, compare solution options, expose risky assumptions, decide what to test next, and preserve the reasoning when new learning changes the decision.
- No login. Working state saves in your browser.
- No synthetic winner score. Decision lenses stay inspectable.
- Decision-trace exports. Evidence, assumptions, tests, learning, and next steps travel with the tree.
Based on the Opportunity Solution Tree framework developed by Teresa Torres. CraftUp is not affiliated with or endorsed by Teresa Torres or Product Talk; this implementation adds evidence tracking and diagnostics around the core structure.
No login. Saved in this browser.
Structural analytics may record actions such as node type, audit opens and export format. Node text, customer observations, evidence URLs and assumptions are not sent as analytics payloads.
Share privacy: the share link contains your tree data. Anyone with the link can inspect it. Compression is not encryption.
100%
OutcomeOpportunitySolutionAssumptionAssumption test
Drag reorders siblings. ↑/↓ buttons provide a keyboard-accessible alternative. Enter adds the default child; Ctrl/Cmd+Enter adds a sibling.
Outcome
Discovery Trace Audit
Specific findings about evidence, framing, divergence, assumptions, learning and structure. No global readiness score.
Outcome
Opportunity space
Loading local tree…
A discovery tree you can audit, not just draw
Step 2
Choose a target opportunity by comparing evidence and decision lenses, then record why the team chose that branch.
Step 3
Explore solution alternatives, expose the assumptions each solution depends on, test the riskiest beliefs, and record what changed.
What belongs on each level?
- Desired outcome
- A product or business result the team wants to influence. Baseline, target, and timeframe can remain unknown until genuinely known.
- Opportunity space
- Customer needs, pains, and desires. The builder tracks whether each branch came from a direct story, analytics, support/sales signal, survey, internal assumption, or another source.
- Solution space
- Several candidate ways to address the selected opportunity when the risk warrants comparison.
- Assumption tests
- Small tests of specific beliefs a solution depends on. CraftUp makes the assumption explicit before its test so the learning object stays visible.
Opportunity or solution?
Review: “Add SSO”
That already specifies an implementation. Ask what customer need makes SSO useful.
Stronger: “I cannot invite my enterprise teammates without security approval”
This leaves room for more than one solution and preserves the customer context.
The built-in Discovery Trace Audit flags likely solution leakage as a review prompt, not as a certainty. It never rewrites or deletes the user's branch automatically.
How target selection works here
CraftUp does not calculate an “OpportunityScore” and declare a winner. Compare customer importance, frequency or reach, evidence strength, satisfaction with current alternatives, strategic relevance, and likely outcome impact independently. Then the team selects a target and records why now, the alternative considered, and unresolved uncertainty.
That preserves the judgment instead of hiding it behind multiplication.
Evidence provenance matters
A direct customer story carries context that an analytics drop or sales objection may not. Analytics, support, success, sales, and surveys remain useful signals, but the tool keeps those source types visible and warns when a branch needs richer customer context instead of calling the opportunity invalid.
Assumption test, not whole-solution theater
Write the belief that must be true first. Then choose a suitable test method, define the observable decision signal up front, and record result, learning, and next decision. Qualitative prototype tests, feasibility spikes, and data checks do not need fake statistical-significance fields.
What the audit is designed to catch
The tree becomes a feature backlog.
Why: Opportunity nodes describe implementations.
Fix: Rewrite them as customer needs, pains, or desires and attach provenance.
A score ends the debate.
Why: Subjective judgments were multiplied into false precision.
Fix: Compare lenses independently and let the team record the target rationale.
Analytics is treated as the customer's need.
Why: A behavioral signal is missing the story/context behind it.
Fix: Keep the signal, label it correctly, and gather customer context before claiming the opportunity is understood.
There is one predetermined solution.
Why: The team converged before exploring alternatives.
Fix: When risk warrants it, add at least one meaningfully different solution for comparison.
Tests are whole-product mini launches.
Why: The underlying risky belief was never made explicit.
Fix: Create an assumption first, then design the cheapest appropriate test of that assumption.
A completed test changes nothing.
Why: Result and next decision were never recorded.
Fix: Capture what happened, what belief changed, and what the team will do next.
A shared URL contains sensitive research.
Why: Compression was mistaken for privacy or encryption.
Fix: Treat the URL as containing the tree; use a safer internal channel or export when content is sensitive.
Imported data disappears or crashes the tree.
Why: Malformed hierarchy was trusted silently.
Fix: Use structural audit findings to repair orphan parents, invalid transitions, multiple roots, or cycles.
Operating notes
- Keep unknown baseline, target, or evidence values unknown rather than manufacturing precision.
- Attach customer-story context to opportunities; use analytics, support, and sales as signals that may need follow-up research.
- If an opportunity sounds like an implementation, ask what customer need would make that implementation valuable.
- Compare opportunity lenses side by side; do not collapse subjective inputs into one synthetic winner score.
- Record why the team selected the target branch and what meaningful alternative it considered.
- Explore more than one solution when the decision is risky enough to justify divergence.
- Make hidden solution beliefs explicit as desirability, usability, feasibility, viability, or ethical assumptions.
- Test the riskiest assumption, not the whole solution, and define the observable decision signal before running the test.
- Record result, learning, and next decision so the tree remains a living discovery artifact.
- Use Markdown or JSON for durable decision history; remember that a share URL contains the tree data itself.
Next step depends on what the tree is missing
Customer evidence is missing
Interview Script GeneratorPrepare story-first discovery questions instead of inventing opportunities inside the tree.
Research notes are messy
Insight Clustering HelperSynthesize evidence before distilling it into opportunity branches.
A risky assumption is clear
Stay in this OSTDesign a lightweight prototype test, one-question survey, data-mining check, research spike, or another appropriate test.
You need a controlled product experiment
A/B Test Plan GeneratorMove to detailed experiment planning only when the test really is a controlled product experiment.
FAQ
Is this Opportunity Solution Tree Builder free?
Yes. The working builder is available without login. It stores the current tree in browser storage, supports exports, and can create a share URL containing a compressed copy of the tree.
Is this an official Teresa Torres or Product Talk tool?
No. CraftUp is not affiliated with or endorsed by Teresa Torres or Product Talk. The builder uses the Opportunity Solution Tree structure developed by Teresa Torres and adds CraftUp-specific evidence tracking, diagnostics, decision rationale, and learning trace.
What are the levels of the tree?
The core flow is desired outcome, opportunity space, solution space, and assumption tests. In this builder, assumptions are explicit objects between a solution and its test so the belief being evaluated never disappears inside a generic experiment node.
Why does the tool not rank opportunities with one score?
Opportunity selection combines messy judgments such as size, customer importance, current satisfaction, strategic relevance, evidence strength, and likely impact on the desired outcome. The tool keeps these lenses inspectable instead of multiplying them into a synthetic answer.
What counts as opportunity evidence?
Direct customer stories provide the richest context for an OST opportunity. Analytics, support, success, sales, and surveys can be useful signals, but the builder labels them separately so they are not silently treated as equivalent to a customer need heard in context.
What if my opportunity only comes from analytics or support?
Keep it. The audit will mark it as needing customer context rather than declaring it invalid. Use the signal to guide follow-up discovery, then update the provenance when you learn more.
What is an assumption test here?
It is a small activity designed to evaluate a specific belief that must be true for a solution to succeed. The builder supports prototype tests, one-question surveys, data mining, research spikes, and other user-defined methods without forcing every discovery question into an A/B test.
Can I record an inconclusive test?
Yes. A completed test can support the assumption, contradict it, or remain inconclusive. The result, learning, evidence reference, and next decision are separate so uncertainty is preserved rather than rewritten as validation.
Is a share link private?
No. The share link contains compressed tree data in the URL. Compression is not encryption. Anyone who receives the link can inspect the tree, so avoid placing sensitive customer observations or internal strategy in a share URL unless that distribution is appropriate.
What happens to old v1 trees?
The v2 loader migrates legacy nodes instead of wiping them. Old evidence is retained with unknown provenance for review, old experiment nodes become assumption tests, and a clearly labeled placeholder assumption is inserted when the original tree never captured the belief being tested.
Learn the method, then use the tool
The builder teaches only enough to prevent misuse. The dedicated OST guide and Product Discovery course own the broader methodology and cadence.
Product change log
Last updated: 2026-09-16
- Replaced the populated new-tree fallback with a genuinely blank outcome workspace and labeled all sample data as illustrative.
- Added structured evidence provenance, opportunity framing diagnostics, explicit target rationale, solution assumptions, assumption tests, and result/learning state.
- Removed score-driven target selection, upgraded exports/share migration to v2, added structural audit checks, and documented share-link privacy truthfully.