Launch Playbook
Frameworks for niche selection, validation, launch, and go/no-go decisions. Complements the Manifest (philosophy) and STREAM (decision-making) with operational rules.
1. Workflow: 5D Contour
Every product cycle follows five phases:
Discovery → Decision → Design → Delivery → Distribution
| Phase | Output | Time |
|---|---|---|
| Discovery | Niche candidates (S.E.E.D. scored) | 1-2 days |
| Decision | GO/NO-GO via STREAM + Fake-Door | 1 day |
| Design | PRD (passes Definition of Done) | 1 day |
| Delivery | Landing page or working demo | 1-3 days |
| Distribution | Content pack + first traffic | 1-2 days |
Total: 5-9 days from idea to live product with traffic
2. Niche Selection: S.E.E.D.
Score each niche candidate on four dimensions:
| Dimension | Question | Signal |
|---|---|---|
| S — Searchability | Can I rank? | Forums/Reddit in top-10, few fresh giants, no video blocks, weak brands |
| E — Evidence | Do people actually have this pain? | Real questions/complaints with URLs |
| E — Ease | Can I build MVP in 1-2 days? | Fits my stack, no external dependencies |
| D — Demand | Will they pay? | Long-tail keywords, clear monetization path |
Kill Flags (stop before starting)
- Top-10 dominated by media giants or encyclopedias
- Fresh guides (<60 days old) already covering the topic
- Video blocks on first SERP positions
- No evidence of real pain (only hypothetical)
- MVP needs more than 1 week on my stack
3. Day 0 Validation: Fake-Door Test
Before writing any code:
- Build a landing page stub (2 hours max)
- Spend $20 on targeted traffic (small ads or relevant communities)
- Measure CTA clicks
Decision:
- < 10 clicks on CTA per $20 → KILL (no oxygen)
- 10+ clicks → proceed to PRD
Goal: cheapest proof that someone cares.
4. PRD Definition of Done
A PRD is not done until it passes this checklist.
- [ ] Problem statement ≥ 30 words (who suffers, when, why now)
- [ ] ICP + JTBD: target segment + 2-3 jobs-to-be-done
- [ ] 3-5 features: each with measurable acceptance criteria
- [ ] 3-5 KPIs: with units (daily/weekly), target values ≥ 0
- [ ] 3-5 risks: with mitigation plan (legal/data/channels/deps)
- [ ] Tech requirements: ≥3 API endpoints, integrations, NFR (speed, logging, limits)
- [ ] Evidence-first: every number/claim has a source URL
Quality Controls
- Evidence-linter: catches numbers without sources, invalid KPIs, malformed endpoints
- Pre-mortem (Counter-PRD): structured document asking “where will this break and how will we notice?”
- Shadow-PRD (optional): alternative pain/UVP formulation for A/B testing the positioning
5. Weekly Decision Rules: Kill / Iterate / Scale
| Week | Metric | KILL | ITERATE | SCALE |
|---|---|---|---|---|
| Week 1 | Fake-Door clicks | < 10 per $20 | — | — |
| Week 2 | Signups | 0 | 1-4 | ≥ 5 |
| Week 3+ | Signup growth WoW | < 0% | 0-50% | > 50% |
ARCHIVE (soft kill): declining signups for 2+ weeks. SCALE actions: new locale, partnerships, programmatic SEO.
6. Delivery: One-Screen Landing
Every product launches with:
- One screen: PROBLEM → DEMO → CTA
- SEO minimum: title, OG tags, JSON-LD, sitemap, hreflang
- A/B headlines: edge-split (Cloudflare Workers or similar)
- Events: view → click → signup/checkout
- P-SEO clusters (optional): /examples, /compare (2-3 subpages)
7. Distribution: Content Pack
Minimum launch content for every product:
- Vertical video (30-45s): HOOK → PROBLEM → DEMO → CTA
- LinkedIn/Twitter post: before/after story, 1 concrete example, CTA
- Community engagement: substantive answers in 2-3 relevant threads
- i18n: EN first + 1 locale where SERP gap is highest
8. Metrics
North Star
Completed “value outcomes” per week per active user.
Weekly Dashboard
| Metric | Description |
|---|---|
| Activation L7/L14 | New user → first value event |
| Cohort retention | Week-over-week retention |
| Payback period | Months to recoup per-customer cost |
| Gross margin | Revenue minus cost-to-serve |
| Active experiments | Count + ROE (result-over-effort) |
Relative Benchmarking
Skip arbitrary “norms.” Compare to your niche:
relative_ctr = your_CTR / niche_benchmark_CTR
relative_conv = your_Conv / niche_benchmark_Conv
score = average(relative_ctr, relative_conv)
Target: ≥ 1.0× niche benchmark.
9. Cadence
Sprint Format (pick one)
- 14-day: Week 1 = research/PRD/prototype, Week 2 = launch/traffic
- Parallel tracks: one product launching, next one in research
Weekly Rituals
| Day | Duration | Focus |
|---|---|---|
| Mon | 90 min | Pick 1-2 decisive steps for the week, freeze everything else |
| Wed | 30 min | Checkpoint: cut anything not moving North Star |
| Fri | 60 min | Retro: what got stronger? what to safely break next? |
Daily Structure
- 2 × 90 min deep work
- 2 × 45 min operations
- 1 slot community engagement
10. Scale Through Leverage
- Find partners who already reach your target audience (platforms, agencies, niche leaders)
- Ship integrations instead of running ad campaigns
- Case studies as currency: 2-3 measurable results open doors
- Start concierge: handle non-critical tasks manually until first revenue
11. Founder-Bandwidth Check
Before committing to a PRD, estimate hours-to-MVP:
| Component | Typical Time |
|---|---|
| UI (if needed) | 2-8 hours |
| Backend/API | 2-4 hours |
| Data/ML pipeline | 4-8 hours |
| Landing page | 1-2 hours |
| Distribution content | 2-3 hours |
If total > 40 hours, you’re bloating “one pain, one function” into a platform. Simplify.
Extracted from startup-agent v1-v4 evolution. No personalities, just operational frameworks.