Handbook
Stage-Gate
Stage-Gate is a structured product development framework that divides the innovation process into stages (where work happens) separated by gates (where decisions happen). At each gate, a cross-functional team reviews…
What it is
Stage-Gate is a structured product development framework that divides the innovation process into stages (where work happens) separated by gates (where decisions happen). At each gate, a cross-functional team reviews evidence and makes an explicit go / kill / pivot / recycle decision before allowing the initiative to proceed.
Originally developed by Robert Cooper in the 1980s for physical product development, Stage-Gate has evolved through five generations. The 5th-generation model (current) integrates Agile practices, AI-assisted decision-making, and adaptive gates — it is not the rigid, waterfall process of earlier versions.
Companies using modern Stage-Gate achieve 63–78% product success rates, compared to ~24% for ad-hoc approaches (PDMA benchmarking studies).
Authoritative sources (external)
| Resource | Executive summary (why it's linked here) |
|---|---|
| Stage-Gate International | Official Stage-Gate body of knowledge — Robert Cooper's framework, 5th-generation updates, research, and case studies. The authority for gate criteria and stage definitions. |
| 5th-Generation Stage-Gate Model | Current evolution of Stage-Gate: iterative stages, Agile integration, lean gates, adaptive governance — addresses historical criticisms of rigidity. |
| PDMA — Stage-Gate 2026 | Product Development and Management Association — practitioner community, research, and updated Stage-Gate guidance. Hosts Cooper's latest articles. |
| ProductPlan — Product development cycle | Glossary overview of the product development cycle — useful when comparing vendor phase counts and labels to Forge P1–P6. |
| Stage-Gate — Discovery-to-launch process | Official discovery-to-launch framing from Stage-Gate International — aligns with gated NPD and complements this blueprint’s mapping table. |
| Atlassian — Product development | Practitioner summary of product development in Agile delivery — how discovery and delivery connect in software teams. |
Core structure
Stages and gates
The classic 5-stage model, adapted for software products:
| Stage | Activities | Key deliverable |
|---|---|---|
| Scoping | Quick assessment: market, technical, financial, strategic fit | One-page opportunity brief |
| Build Business Case | Detailed investigation: user research, competitive analysis, concept testing, feasibility | Business case with evidence |
| Development | Design and build the product (via SDLC) | Working increment |
| Testing & Validation | Market testing, beta, usability, operational readiness | Validated product ready for launch |
| Launch | Market activation, GTM execution, scaling | Product in market |
Gate decisions
Each gate uses explicit criteria — not opinions:
| Decision | Meaning | Action |
|---|---|---|
| Go | Evidence meets criteria; proceed to next stage | Fund next stage, assign resources |
| Kill | Evidence shows the initiative should not proceed | Stop work, reallocate resources, document learnings |
| Pivot | Evidence suggests a different direction is more promising | Recycle to an earlier stage with revised framing |
| Hold | Insufficient evidence to decide; more information needed | Time-boxed investigation, then re-gate |
5th-generation adaptations
| Evolution | What changed |
|---|---|
| Agile integration | Stages use Agile/Scrum for execution; gates remain as strategic decision points |
| Lean gates | Gates are lighter, data-driven, and faster — not bureaucratic review boards |
| Spiral / iterative stages | Stages iterate internally (build-test-learn) rather than being strictly sequential |
| Adaptive | Gate criteria adjust based on project risk and novelty — higher risk = more rigorous gates |
| AI-assisted | AI can prepare gate materials, analyze market data, and flag risks — humans make decisions |
Discovery-to-launch reference framing (seven stages)
There is no single universal PDLC phase model. Teams differ on how many phases they use, where one phase ends and the next begins, and whether the lifecycle stops at launch or continues into post-launch learning. Strong practice is to define each phase by purpose, required activities, deliverables, exit criteria, and gate decisions, while treating the flow as iterative and cross-functional, not as a rigid one-way handoff. (See also ProductPlan — product development cycle.)
The table below is a mainstream seven-stage discovery-to-launch shape (similar to Stage-Gate’s discovery-to-launch material), with one distinct decision question per stage — a quick sanity check that each stage earns its place. Forge’s canonical lifecycle remains P1–P6 plus embedded SDLC A–F in Product development lifecycle (PDLC) and PDLC ↔ SDLC bridge; use the mapping column to translate vocabulary.
| Stage | Distinct decision question | Typical Forge mapping |
|---|---|---|
| 1. Discovery | Is this a real opportunity? | P1 Discover Problem |
| 2. Scoping / screening | Is it promising enough to justify deeper work? | P1 → P2 (lightweight; may align with early G1) |
| 3. Business case / planning | Is it worth funding? | P2 + P3 (validation + commit; G2, G3) |
| 4. Design / prototype | Do we know what to build with enough confidence? | Late P2 + handoff into SDLC A–C |
| 5. Development | Can we build it to the agreed bar? | SDLC D (and supporting C) |
| 6. Test and validate | Does it work in reality (users, ops, market)? | SDLC E–F + P4 (beta / readiness) |
| 7. Launch and learn | Can we win with it, operate it, and learn? | P4 Launch + P5 Grow (G4, G5) |
Merging or splitting phases is normal. For example, some teams merge design into development; others treat post-launch learning as its own emphasis inside P5. What matters is that each phase still has a clear purpose, explicit deliverables, and exit criteria — not the label count.
Mapping to PDLC phases
Stage-Gate maps directly to PDLC phases and gates:
| Stage-Gate | PDLC phase | PDLC gate |
|---|---|---|
| Gate 0: Idea Screen | — | Pre-P1 screening |
| Stage 1: Scoping | P1 Discover Problem | — |
| Gate 1: Second Screen | — | G1 (problem worth solving?) |
| Stage 2: Business Case | P2 Validate Solution + P3 Plan & Commit | — |
| Gate 2: Go to Development | — | G2 (solution viable?) + G3 (invest in building?) |
| Stage 3: Development | SDLC A–F | — |
| Gate 3: Go to Testing | — | (SDLC Phase E exit) |
| Stage 4: Testing & Validation | SDLC E–F + P4 Launch (beta) | — |
| Gate 4: Go to Launch | — | G4 (ready to launch?) |
| Stage 5: Launch | P4 Launch (GA) | — |
| Post-Launch Review | P5 Grow | G5 (continue investing?) |
Using Stage-Gate within this blueprint
Use Stage-gate review — [Initiative Name] to document gate decisions. The template captures: - Gate identifier and date - Evidence reviewed (with links to artifacts) - Criteria assessment (met / not met / partially met) - Decision (go / kill / pivot / hold) - Conditions for proceeding (if any) - Resource allocation for next stage
Anti-patterns
| Anti-pattern | Fix |
|---|---|
| Gates as bureaucratic checkpoints | Gates should be fast, evidence-based decisions — not multi-day review boards. If gates slow you down, make them leaner, not fewer. |
| Skipping gates for "urgent" projects | The riskier the project, the more you need gates. "Urgency" is not evidence of viability. Adapt gate rigor, don't skip. |
| Gates without kill authority | If a gate never kills a project, it's theater. Empower gatekeepers to say no. Track kill rate as a health metric. |
| Waterfall stages | Use Agile execution within stages. Stage-Gate provides strategic decision points; stages are iterative, not sequential waterfalls. |
Further reading
- PDLC.md §4 — Stage gates — Gate definitions G1–G5 in this blueprint
- PDLC-SDLC Bridge §6 — Decision framework — Meeting questions, when to use heavy vs light gates, plain-language gates vs G1–G5
- Lean Startup — Complementary: hypothesis-driven validation within stages
- Design Thinking — Complementary: human-centered methods for Stages 1–2