Plan funnel

Funnel Planner — Orchestrator

Strategy — Step 3 of 4. Numeric funnel + unit-economics model: rates, traffic, sensitivity bands. Drives downstream channel-mix + budget decisions.

Core Question: "Do the numbers actually work?"

Philosophy, improvement-factor calibration, 9-channel reference, growth-motion ID, when NOT to use: references/playbook.md [PLAYBOOK].

---

Critical Gates — Read First

Six hard gates (fire under --fast too — full detail in references/procedures/critical-gates.md [PROCEDURE]):

1. Numeric baseline on every target — zero TBDs; benchmarks OK with confidence flag.
2. Justification cited — baseline + improvement factor + reasoning, no naked numbers.
3. 70% test — partial achievement must still be valuable (higher/lower/binary rules).
4. LTV:CAC ≥ 3:1 for acquisition, else explicit Pricing Health Signals flag.
5. Growth motion identified — PLG / SLG / Hybrid drives model + channel-mix.
6. Three-outcome validation — Business (required), Brand, Community; N/A needs justification.

Quality Gate

Critic 4-point rubric (pass-all): numeric baselines · justified targets · 70% test · LTV:CAC. FAIL → rewrite loop, max 2 cycles, then done_with_concerns with failures pinned as Known Issues.

---

Before Starting

Per references/_shared/before-starting-check.md [PLAYBOOK]:

Artifact	Source	Required?
docs/forsvn/artifacts/meta/sketches/prioritize-*.md	prioritize	Hard gate — missing → NEEDS_CONTEXT
Prior targets-*.md (same initiative)	self	Optional — enables Route C delta
.forsvn/index/manifest.json	manifest-sync	Recommended — >30d warns
Industry baselines	references/benchmarks.md	Fallback when user lacks rates

Mode (standard): --fast forces Route C if conditions hold, else Route B (no critic). Hard gates still enforced. references/_shared/mode-resolver.md [PROCEDURE]. Pre-Dispatch (hard-gate, read order, Cold/Warm Start, staleness, growth-motion ID, write-back): references/procedures/pre-dispatch.md [PROCEDURE]. Session execution profile (single-vs-multi): inherit per references/_shared/execution-policy.md.

---

Artifact Contract

• Path: docs/forsvn/artifacts/meta-plan-funnel-<YYYY-MM-DD>-targets-<slug>.md (flat v2; re-run renames prior .v[N])
• Lifecycle: snapshot · review_surface: none
• Frontmatter: skill, version, date, status, stack (=meta), review_surface (=none)
• Required sections (cross-stack contract): Funnel Stages · Target Table · Channel → Funnel Stage Map · Three-Outcome Validation · Validation · Baselines
• Side effects: append baselines + growth motion + unit economics to experience/business.md (mandatory on PASS / done_with_concerns)
• Consumed by: plan-campaign, run-pipeline, future re-runs (delta detection)

Full schema: references/procedures/contract.md + references/format-conventions.md [PROCEDURE].

---

Agents + Dispatch

6 sub-agents (full focus in agents/):

• L1 parallel: model-selection (model + stage map + channel-to-stage), baseline-collector (rates, benchmarks, unit-economics)
• L2 sequential: target-setter (targets + improvement factors + sensitivity bands) → sanity-check (6-pattern subset) → stress-test (revenue, 70%, ownership, measurement) → critic (4-point Quality Gate, max 2 rewrite cycles)

Three routes — A Full (--deep OR 3+ initiatives across 2+ models: L1 parallel → merge → L2 seq) · B Standard default (known model, 1-2 initiatives: target → sanity → critic) · C Fast Bump (--fast + ≤3 sentences + single initiative + prior targets-*.md: target-setter only).

Spawn mechanics, merge, critic FAIL routing, fallbacks, side effects, chain position: references/procedures/dispatch-mechanics.md [PROCEDURE].

---

Anti-Patterns

references/anti-patterns.md [ANTI-PATTERN] — 10-pattern catalog (vanity metrics, feature factory, sandbagging, aspirational math, orphan owners, input traps, unit-economics blind spots, cross-stack contract drift, +2). Sanity-check scans 6; critic reads all. Re-read before ship.

---

Durable Rules (protected)

<!-- SLOW_UPDATE_START --> <!-- No pinned rules yet. Populate via the slow-update workflow (see references/slow-update-fence.md). Each pinned rule must (a) be procedural not instance-specific, (b) be earned from a regression or critic-flagged failure, (c) cite the artifact / decision record that justified pinning. --> <!-- SLOW_UPDATE_END -->

Completion Status

• DONE — model selected, baselines collected, targets stress-tested, critic PASS
• DONE_WITH_CONCERNS — targets computed but baselines thin/estimated, or critic loop cap hit (failures pinned as Known Issues)
• BLOCKED — no baseline data AND LTV:CAC <3:1; needs business-model fix first (defer to diagnose)
• NEEDS_CONTEXT — prioritize-*.md missing OR baselines absent; recommend prioritize (hard-gated, no INTERVIEW substitute)

Next Step

On PASS, hand the artifact to plan-campaign for channel-mix execution. On drift or new initiatives, re-run for delta detection.

---

References

• Worked example: references/examples/funnel-planner-walkthrough.md [EXAMPLE]
• Procedures: references/procedures/{critical-gates, contract, pre-dispatch, dispatch-mechanics}.md
• Shared: references/_shared/{before-starting-check, mode-resolver, pre-dispatch-protocol, artifact-contract-template}.md
• Data catalogs: funnel-models.md, benchmarks.md, unit-economics.md, stress-tests.md, format-conventions.md
• Agents in agents/; critic-agent.md holds the canonical rubric. Stack conventions: research-skills/CLAUDE.md.