borghei/Claude-Skills / project-management/discovery/opportunity-solution-tree
opportunity-solution-tree
VerifiedMaps one outcome to customer opportunities, 3+ solutions each and assumption tests, then audits the tree. Use when prioritizing discovery work, mapping solutions to a problem, or checking whether a roadmap moves outcomes.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 9.3 KB |
| assets/ost_template.md | markdown | 2.0 KB |
| assets/sample_ost.json | config | 3.5 KB |
| assets/sample_ost_flawed.json | config | 609 B |
| references/ost-anti-patterns.md | markdown | 4.6 KB |
| references/ost-fundamentals.md | markdown | 6.4 KB |
| scripts/ost_validator.py | script | 10.7 KB |
references/ost-fundamentals.md
View on GitHub# OST — Fundamentals Reference
From Teresa Torres' *Continuous Discovery Habits*.
## 1. Why the OST exists
Most product teams jump from problem to solution. The OST forces a
deliberate exploration of multiple opportunities and multiple solutions
before committing.
Common failure mode without OST:
1. Problem identified
2. First solution proposed
3. Solution accepted
4. Build
5. Ship
6. Outcome doesn't move
7. Repeat
The OST surfaces the divergent thinking that prevents step 6.
## 2. The 4 layers
### Layer 1: Outcome
One specific, measurable target.
Discipline:
- ONE outcome per tree (not 3 outcomes)
- Behavioral or business (not output)
- Bounded timebox (this quarter / half)
- Team can influence (not "increase revenue" for an eng team)
### Layer 2: Opportunities
Customer needs, pains, desires.
Discipline:
- Written from customer point of view ("users can't X")
- Grounded in evidence (which interview / ticket / data point?)
- Not solutions
- 3-7 distinct opportunities per outcome (more = ineffective)
### Layer 3: Solutions
Candidate ways to address each opportunity.
Discipline:
- Multiple per opportunity (3-5 ideal)
- Not just one obvious one
- Varied in cost / effort / risk
- Don't promote to roadmap until tested
### Layer 4: Assumption tests
The experiments that validate or invalidate solutions.
Discipline:
- Cheapest test that gives confidence
- Test the riskiest assumption first
- Pre-define success criteria
- Time-bounded
## 3. Outcome examples
### Good outcomes
- "Lift week-1 activation from 28% → 40% by Q3"
- "Reduce admin time-on-task for invoicing from 18min → 8min"
- "Increase free-to-paid conversion in segment X from 4% → 7%"
- "Achieve NPS ≥ 55 from enterprise users (currently 38)"
### Bad outcomes
- "Improve user experience" — vague, unmeasurable
- "Build a new dashboard" — output
- "Hit $50M ARR" — company-level, not team-level
- "Be the best in our market" — aspirational
- "Make customers happy" — undefined
## 4. Opportunity examples
### Good opportunities
- "Trial users abandon during email-verification step (35% drop-off)"
- "Admins want to bulk-invite users from CSV (mentioned in 6 of 8 interviews)"
- "Customer-success teams can't find churn-risk customers fast enough"
- "Enterprise procurement blocks deals without SOC 2 evidence"
### Bad opportunities
- "Need to add CSV import" — solution
- "Users want better dashboard" — too vague
- "Improve email flow" — too vague
- "Customers want AI" — too generic
## 5. Solution examples
For opportunity "Admins want to bulk-invite users from CSV":
### Good solution alternatives
1. CSV upload UI with templated download
2. API endpoint for invite via existing tools (Okta, Workday)
3. Salesforce-style "import wizard" with field mapping
4. Office365/Google Workspace sync (auto-invite based on group)
5. SCIM / directory sync (enterprise-grade)
Each solves the same opportunity differently. Cost, effort, segment fit differ.
### Bad solution
1. "Build a feature for bulk invites"
Single, vague, no exploration.
## 6. Assumption test examples
Solution: "CSV upload UI with templated download"
### Assumption categories
- **Value:** "Admins will actually use this if available"
- **Usability:** "Admins can complete bulk invite in <5 min without help"
- **Feasibility:** "We can build it in 2 weeks"
- **Viability:** "Reduces support tickets, increases activation rate"
### Test the riskiest
Risk + evidence quadrant:
- High risk + low evidence = TEST NOW
- High risk + high evidence = ship; monitor
- Low risk + low evidence = test cheaply
- Low risk + high evidence = no test needed
### Cheap tests
- Value: 5 admin interviews; ask "have you ever wanted to bulk invite? what did you do?"
- Usability: paper prototype with 5 admins
- Feasibility: 1-day eng spike
- Viability: estimate support ticket reduction × admin time saved
## 7. The weekly OST rhythm
Teresa Torres recommends weekly interviews:
### Cadence
- 1+ customer interview per week
- 1 OST update per week (15-30 min team review)
- 1 assumption test in flight at any time
- Quarterly tree refresh (kill stale branches)
### Without weekly rhythm
- OST becomes stale
- Discovery becomes episodic
- Team falls back to feature factory
### Roles
- PM owns the OST
- Designer + engineer collaborate
- Researcher (if you have one) drives interviews
- Whole team reviews weekly
## 8. OST and roadmap
The OST informs the roadmap; the roadmap doesn't replace the OST.
- OST captures: outcome, opportunities, solutions, tests
- Roadmap captures: what we're committing to ship, when, with what risk
Solutions that pass assumption tests → graduated to roadmap.
Solutions still in test → stay in OST.
Solutions that failed tests → killed; opportunities remain (with new candidates).
## 9. Multiple OSTs across a team
A team might have 2-3 OSTs going if working on 2-3 outcomes. More than
3 = trying to do too much.
A company has multiple OSTs across teams, all rolling up to company OKRs.
## 10. Common Torres principles applied
- **Continuous, not episodic.** Discovery happens weekly, not quarterly.
- **Outcomes over outputs.** Measure success by behavior change, not by shipped features.
- **Multiple opportunities, multiple solutions.** Diverge before converging.
- **Test cheaply.** Cheapest test that resolves the highest risk.
- **Whole-team discovery.** PM + designer + engineer all participate.
- **Visualize.** The tree itself is the artifact.
## 11. Tools
Common OST tools:
- Miro / Mural (most popular for the visual layout)
- FigJam
- Whimsical
- Lucidchart
- Notion (text version)
- Spreadsheet (text version)
Tool < discipline. A spreadsheet-OST updated weekly > a Miro masterpiece updated quarterly.
## 12. When the OST gets stuck
### Symptom: no progress in a quarter
- Outcome unclear / unmeasurable
- Solutions chosen without testing
- Tree not updated
**Fix:** Re-check outcome; commit to one assumption test per week.
### Symptom: too many opportunities
- Trying to cover too much
- Outcome is too high-level
**Fix:** Decompose outcome to smaller; pick top 3 opportunities.
### Symptom: keeps coming back to same solution
- Team has bias / pre-decided
- Brainstorm isn't divergent enough
**Fix:** Bring outside perspective. Borrow from analogous problems.
### Symptom: assumption tests aren't run
- "Too busy" — actually means deprioritized
- Tests overdesigned (waiting for perfect)
**Fix:** Time-box test to 1 week max. Cheapest valid test.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |