Research Assessment
Systematically evaluate research artifacts to determine production readiness.
When to Use
- Evaluating a POC before committing to production build
- Assessing what an experiment actually proved vs assumed
- Identifying gaps between prototype and production requirements
- Creating readiness checklists for stakeholder review
- Analyzing ADRs for implementation implications
Core Framework: "What Did We Prove?"
Before planning production work, separate fact from assumption:
Validated (Evidence Exists)
Claims with concrete evidence:
- Test results that passed
- Benchmarks with data
- User feedback collected
- Integration confirmed working
Assumed (No Direct Test)
Claims we believe but haven't verified:
- "It should scale" (but no load test)
- "Users will want this" (but no validation)
- "Security is fine" (but no audit)
Unknown (Gap)
Things we don't know:
- Untested edge cases
- Unexplored failure modes
- Missing requirements
Production Readiness Checklist
Score each criterion 1-10:
| Criterion | Weight | Questions to Ask |
|---|---|---|
| Core hypothesis validated | 20% | Did the experiment prove what we set out to prove? |
| Performance benchmarks | 15% | Do we have latency, throughput, resource usage data? |
| Security considerations | 15% | Have we identified and addressed security risks? |
| Scalability tested | 10% | Will it work at 10x, 100x current scale? |
| Error handling | 10% | What happens when things fail? |
| Observability | 10% | Can we monitor and debug in production? |
| Documentation | 10% | Could someone else take this over? |
| Knowledge transfer | 10% | Is knowledge spread across the team? |
Scoring Guide:
- 8-10: Production-ready with minor polish
- 6-7: Needs targeted work in specific areas
- 4-5: Significant gaps requiring substantial effort
- 1-3: Research phase, not ready for production planning
Gap Analysis Matrix
Use this template to map what's proven vs unknown:
| Domain | Proven | Assumed | Unknown |
|---|---|---|---|
| Functionality | Feature X works | Feature Y will work similarly | Edge case behavior |
| Performance | 100ms p50 latency | Will scale linearly | Performance under load |
| Security | Auth works | No vulnerabilities | Penetration test results |
| Scalability | Works for 100 users | Will work for 10K | Actual breaking point |
| Operations | Can deploy manually | Automation will work | Rollback procedure |
| Integration | API contract defined | Will integrate smoothly | Error handling at boundaries |
Red Flags Checklist
Watch for these patterns that indicate low readiness:
- Happy path only - no error handling
- Hardcoded values that need configuration
- "TODO" comments for critical functionality
- No tests or only manual testing
- Single contributor (bus factor = 1)
- Missing logging/monitoring hooks
- No documentation of design decisions
- Untested third-party dependencies
- Security considerations deferred
- No rollback or recovery plan
Output Template
# Research Assessment: [Project Name]
## Executive Summary
[2-3 sentences: what this is, key finding, readiness score]
## What Did We Prove?
### Validated Hypotheses
- **[Hypothesis]**: [Evidence] (Confidence: High/Medium/Low)
### Key Findings
1. [Finding with evidence]
## Production Readiness: X/10
| Criterion | Score | Notes |
|-----------|-------|-------|
| Core hypothesis | X | |
| Performance | X | |
| Security | X | |
| Scalability | X | |
| Error handling | X | |
| Observability | X | |
| Documentation | X | |
| Knowledge transfer | X | |
## Gap Analysis
| Domain | Proven | Assumed | Unknown |
|--------|--------|---------|---------|
| ... | ... | ... | ... |
## Risks & Technical Debt
- [Item with severity]
## Recommended Next Steps
1. [Action]
Progressive Validation Ramp
Before investing in production readiness, projects must pass increasingly rigorous gates. Apply the "80% of experiments should fail" philosophy:
| Gate | Kill Rate | Key Question |
|---|---|---|
| Idea → Internal Build | ~40% | Is the problem real and worth solving? |
| Internal → Private Preview | ~30% | Have we built something worth testing externally? |
| Private → Public Preview | ~30% | Are external users getting value? |
| Public → GA | ~10% | Are we ready to bet the company reputation? |
See references/graduation-criteria.md for detailed gate requirements.
References
See references/ for:
readiness-criteria.md- Detailed scoring rubricscommon-gaps.md- Typical gaps by project typegraduation-criteria.md- Gate criteria and kill decisions (GitHub Next patterns)rfc-templates.md- Google, Uber, Sourcegraph RFC formats
See examples/ for:
poc-assessment-example.md- Vector Search POC with full "What Did We Prove?" frameworkgate-review-example.md- Internal Build → Private Preview graduation checklist