/nexus

Growth-Acceptance Blueprint — the layer=c preset of acceptance

/nexus growth-acceptance ≡ /nexus acceptance layer=c — the lifecycle-layer preset of the acceptance merge-adjudication engine. growth-acceptance is a named alias kept for discoverability; merge-time adjudication (Phases 0-6, Layer A/B, G1-G10, joint verdict) is owned by reference/acceptance-recipe.md and is not re-derived here — Phase 1 below delegates to it wholly at Tier B. This file owns only what the layer adds: the pre-design insight contract, ship-time market/brand/regulatory setup, the post-launch measurement loop with auto-halt, and G11-G15.

Recipe contract for nexus growth-acceptance — orchestrates the lifecycle gates (pre-design → ship-time → post-launch) defined by _common/GROWTH_BRAND_PROOF.md (Layer C: Market + Research + Brand axes). Use when an Enterprise-tier organization wants AI-generated content / campaigns / brand-touching artifacts to be machine-adjudicated across their entire lifecycle, not only at merge.

Prerequisites:

• _common/PROOF_CARRYING.md v3 (mandatory — Tier A Foundation + Tier B Production Pipeline + cross-cutting G11/G14/G15)
• _common/GROWTH_BRAND_PROOF.md (mandatory — Layer C protocol)
• nexus/reference/acceptance-recipe.md (Phase 1 delegates here for merge-time gates)

Distinguishes from / integrates with:

• acceptance — Code + Design merge-time gate (Tier B); growth-acceptance includes it as Phase 1 (full delegation, shared v3 protocol foundation) and extends to pre-design (Phase 0) + ship-time (Phase 2) + post-launch (Phase 3).
• Every other distinction acceptance draws (feature, apex, summit, judge) applies here unchanged — see reference/acceptance-recipe.md § Distinguishes from. The Layer-C-specific chaining on top: apex authors → growth-acceptance gates the lifecycle; summit decides → growth-acceptance gates execution; kaizen produces → growth-acceptance gates brand-touching surfaces; and feature is the recommended downgrade when Phase 0 detects Solo / SMB or sub-Step-2 adoption.

When to Invoke

Trigger nexus growth-acceptance when all apply:

• Org Tier is Enterprise (50+ people with Research / Brand / Growth-analytics specialists)
• Step 2+ of Phased Adoption is in place (Insight Ledger operational + Research Lead role staffed)
• The change is brand-touching content / campaign / customer-facing message (not internal infra / pure backend / dev tooling)
• The organization has committed to lifecycle-gating (this is a workflow choice, not a tactic)

Do not invoke when:

• Org Tier is Solo / SMB → use acceptance (Tier B) only; Step 1 Measurement Loop optional opt-in
• Pure backend / infrastructure / data pipeline change → use acceptance Tier B
• Regulated industry where Layer C autonomy is legally restricted → use acceptance Tier B + manual Layer C review per G14
• User wants exploration / brainstorming, not lifecycle gating → use flux / magi / omen
• Layer C Step adoption is below Step 3 → use acceptance Tier B + manual Market / Brand review

Phase Contract

Phase 0 — Pre-Design (Research + Insight + Contract Draft)

Goal: No content authoring begins until evidence + Contract are in place.

Agents (sequential):

1. Nexus[classify] — internal: detect org_tier, step_adoption_level, change_category (content / campaign / brand-touch), regulatory_jurisdiction, tier (S/A/B/C), ui_dimension (none / partial / full)
2. insight (proposed skill) or field — query Insight Ledger for relevant evidence; if no evidence available for the proposed segment, REJECT (block design)
3. field — if Step 2+ and evidence is stale or missing, queue fresh research
4. scribe[unified] + spark — draft Growth-Brand Contract (Tier 0 / 1 / 2 per scope)
5. magi (optional) — high-stakes content: arbitrate Strategic-level Constitution alignment

Gate (G11 mandatory):

• Insight Ledger consultation is mandatory; thinking-aloud claims are rejected
• Contract Tier 1+ requires insight_refs field with valid Ledger IDs
• AI cannot write to Insight Ledger (read-only); proposed-edit queue must clear before design begins

Engine routing: Field / Voice / Trace on Codex for data extraction; Insight Lead synthesis on Claude (judgment); Echo[demand] on Claude for customer empathy.

Output: Approved Contract + Insight Ledger citation set + Constitution Operational-layer rule citations.

Phase 1 — Merge-Time (delegates to acceptance recipe Tier B)

Goal: Code + Design merge-time gates per existing acceptance recipe.

Delegation: This phase fully and wholly delegates to nexus acceptance recipe (nexus/reference/acceptance-recipe.md) — its complete Phase 1-6 chain (Spec Diff → Layer A/B Oracle Generation → Layer A/B Adversarial Exploration → Layer A/B/Joint Acceptance Gate → Runtime Oracle Hookup) runs unmodified; growth-acceptance never re-implements or re-derives any part of that adjudication (see reference/acceptance-recipe.md § Phase Contract for the full agent-level detail).

Tier C extensions during Phase 1:

• Brand Compiler B.hard layer runs during Phase 4B alongside Design Compiler — blocks taboo words / legal compliance / Distinctiveness Floor violations
• Brand Compiler B.pattern layer runs during Phase 4B — blocks token / structure / format compliance failures

Gate: Per acceptance recipe Phase 4C joint verdict. Layer C extensions block merge if B.hard or B.pattern FAIL.

Phase 2 — Ship-Time (Market Setup + Brand Advisory + Regulatory Check)

Goal: Before deploy / launch, verify measurement design + brand advisory + regulatory compliance.

Agents (parallel):

1. pulse + experiment — Market Proof: declare incrementality_proof measurement method per Decision Tree (Conversion Lift / GeoLift / MMM / Synth); compute statistical power
2. ledger (FinOps) — CAC / LTV thresholds; budget allocation
3. compete — cannibalization estimation
4. funnel + funnel[premium] — channel-fit rationale and LP coherence
5. vision + prose — Brand Compiler B.tone advisory layer (LLM-as-judge, non-blocking)
6. canon[legal] + canon[regulatory] + cloak + vigil — G14 Regulatory Envelope Pre-Flight: verify regulatory_jurisdiction toggles, brand-safety placement exclusions

Gate (G14 mandatory):

• Regulatory jurisdiction declared; per-jurisdiction toggle (auto-scale / auto-generate / holdout test) verified
• Regulated industries (medical / financial / political / pharmaceutical): auto-scale OFF default, manual approval required
• Statistical power adequate for declared measurement method (MDE check)
• B.tone advisory findings recorded as unspecifiable_advisory (non-blocking)

Pre-Registration sub-gate (the integrity backbone — freeze the plan before the data exists):

A measurement-backed launch is only honest if the success bar is set before the outcome is known; otherwise the +14/+30/+90d gates can be gamed by post-hoc metric selection (HARKing / p-hacking). Before ship, lock an immutable analysis plan:

• Primary metric (single, pre-declared — the incrementality outcome that decides scale/halt), MDE, decision thresholds (auto-scale / auto-halt cut-offs), analysis window, and the decision rule itself (what value triggers scale vs halt). Secondary metrics are explicitly labelled secondary and cannot promote to primary post-launch.
• The plan is frozen and hash-stamped into the Contract before ship. Phase 3 evaluates against this frozen plan only.
• Control-validity check — experiment verifies the incrementality counterfactual is sound BEFORE launch: holdout/geo-control assignment is valid, SRM (sample-ratio-mismatch) clean, and test/control contamination bounded. A broken control invalidates the incrementality claim no matter what lift is observed — block ship until the control is sound.

Output: Ship-eligible artifact with full Market Proof setup + frozen pre-registered analysis plan (hash-stamped) + validated control + Brand Advisory record + Regulatory clearance.

Phase 3 — Post-Launch (Measurement Loop + Auto-Halt + Learning)

Goal: Continuous measurement after launch; auto-scale on win, auto-halt on loss, brand erosion → emergency review.

Schedule (scheduled gates, not merge-time):

• +14 days — initial lift check (early-signal)
• +30 days — incrementality confirmed; CAC / LTV trend; cannibalization check
• +90 days — Brand Lift / brand search query trend / Distinctive Asset audit; full Learning record

Agents (per scheduled gate):

1. pulse + experiment — collect post-launch metrics per Incrementality method
2. compete — cannibalization analysis (cross-campaign impact)
3. ledger — CAC / LTV trend update
4. beacon — Brand Lift monitoring (if Brand Lift Proof was set up)
5. mend — if Stop_Condition triggered: G13 auto-halt sequence
6. launch[weekly] — aggregate findings into Learning record
7. tome — update Insight Ledger with validated findings (queue for Research Lead merge per G11)

Gate (G13 mandatory):

• Evaluate against the frozen pre-registered plan only — read the primary metric, MDE, and thresholds from the Phase 2 Pre-Registration sub-gate's hash-stamped plan; they are not re-chosen now. Any proposed post-launch change to metric/threshold/window is a flagged deviation requiring explicit Stop_Accountable + Research Lead sign-off (recorded with reason) — never a silent re-baseline.
• Stop_Condition trigger fires → Stop_Accountable notified immediately
• 24h no-response → auto-halt (default deny)
• Brand Director has unilateral veto on Brand-related stops (Growth Lead cannot override)
• Auto-scale eligibility requires: positive incrementality on the pre-registered primary metric with a valid control (SRM clean) + Brand Lift not degraded + CAC < threshold + no cannibalization > X%

Output: Per-checkpoint Learning record + Auto-scale decision (or Auto-halt + post-mortem) + Insight Ledger proposed-edit queue entry.

Phase 4 — Continuous Cross-Cutting Audits (background, not per-change)

These run on schedule, not per-change, but affect every Layer C operation. Audit content and mechanics are defined in _common/GROWTH_BRAND_PROOF.md (§ Cross-cutting Application of G11/G14/G15 and the Ledger/Constitution audit sections); cadence below is this recipe's own schedule:

• Quarterly Constitution Health Audit (G15)
• Quarterly Goodhart Audit (G6 carry-over from Tier B)
• Quarterly Regulatory Horizon Scan (G14)
• Quarterly Distinctive Asset Audit (G12) — Brand Voice Distinctiveness Index; competitor Colour Stealing detection
• Monthly Insight Ledger Audit (G11)
• Monthly Override Audit (G14 + G15 + Hot-Fix Fast-Path)
• Weekly B.tone Sampling Audit (G7 carry-over)

Chain Template (AUTORUN)

Phase 0 (Pre-Design, sequential):
  Nexus[classify org_tier + step_adoption + change_category + regulatory_jurisdiction + tier + ui_dimension]
  → if org_tier=Solo: abort with `acceptance` recommendation
  → if org_tier=SMB AND step_adoption < 2: abort with `acceptance` + Step 1 Measurement Loop recommendation
  → insight[query Ledger for evidence] (read-only per G11)
  → if evidence stale/missing AND Step ≥ 2: field[queue fresh research, block design]
  → scribe[unified] + spark[draft Growth-Brand Contract Tier 0/1/2]
  → if change_category=brand_critical: magi[Constitution alignment]

Phase 1 (Merge-Time, delegates to acceptance recipe):
  nexus acceptance (full chain with Tier B Layer A + Layer B + B.hard/B.pattern extensions)

Phase 2 (Ship-Time, parallel):
  ‖ pulse + experiment[Market Proof setup + Incrementality Decision Tree]
  ‖ ledger[CAC/LTV thresholds]
  ‖ compete[cannibalization estimation]
  ‖ funnel + funnel[premium: channel-fit + LP coherence]
  ‖ vision + prose[B.tone advisory, non-blocking]
  ‖ canon[legal] + canon[regulatory] + cloak + vigil[G14 Regulatory Pre-Flight]
  ‖ experiment[pre-register + control-validity/SRM check, per § Pre-Registration sub-gate]
  → Gate: regulatory toggle verified + statistical power adequate + analysis plan frozen + control valid
  → if FAIL: block ship; route to remediation

Phase 3 (Post-Launch, scheduled):
  +14d:
    ‖ pulse + experiment[initial lift]
    ‖ beacon[early Brand Lift]
  +30d:
    ‖ experiment[incrementality confirmed]
    ‖ ledger[CAC/LTV trend]
    ‖ compete[cannibalization]
  +90d:
    ‖ beacon[Brand Lift]
    ‖ compete[Distinctive Asset audit]
    ‖ launch[Learning record]
    ‖ tome[Insight Ledger proposed-edit queue]
  Per-checkpoint Gate (G13) — evaluated against the frozen plan; rules in § Phase 3:
    Stop_Condition fires → notify Stop_Accountable → 24h no-response → mend[auto-halt]

Phase 4 (Cross-Cutting Audits, background):
  Quarterly: G15 Constitution Health / G6 Goodhart / G14 Horizon Scan / G12 Distinctive Asset
  Monthly: G11 Insight Ledger / Override Audit
  Weekly: B.tone Sampling

Failure Escalation

4 further rows (Insight Ledger AI-proposal rubber-stamping, Brand Voice Distinctiveness Index degradation, Emergency Override-count review, Constitution staleness lockout) are the Phase 4 audit-triggered failure modes — detection detail in _common/GROWTH_BRAND_PROOF.md; escalation actions are as the pattern above (flag / block / process review).

Cost & Scale Profile

Note: Phase 1 cost varies per acceptance recipe Tier (Tier-S full UI 9-15×, Tier-A full UI 5-8×, etc.). Total cost = Phase 1 cost + Tier C overhead from above.

Confirm with user before launching Step 3 or Step 4 — costs and organizational requirements rise sharply. Strongly confirm before Step 4 — Brand Compiler blocking changes the brand-team workflow.

Output Report — Lifecycle Dossier (named)

Extends the engine's Acceptance Dossier (reference/acceptance-recipe.md) rather than replacing it — the merge-time sections come from there unchanged. Layer C adds:

• Insight Ledger — each pre-design insight with its evidence and the contract clause it produced
• Market / Research / Brand axis verdicts — G11-G15 disposition per _common/GROWTH_BRAND_PROOF.md
• Ship-time setup record — market, brand-compiler, and regulatory checks with their outcomes
• Post-launch measurement state — the metric contract, current readings, auto-halt status, and the next scheduled checkpoint (+14/+30/+90d)
• Learning entry — what the launch taught, written back for the next cycle

Confirm / Safety Gate

Inherits the engine's gates unchanged (reference/acceptance-recipe.md) — Confirm before launch when tier == S, plus standard Ask First tiers. Layer C adds one of its own: Ask First before any ship-time action that touches an external surface (market listing, brand asset publication, regulatory filing), since those are outward-facing and not reversible by a revert.

Resume

Schedule-resume — Layer C's post-launch phase is a cadence, not a continuous run: state persists between the +14 / +30 / +90d checkpoints and each wake-up resumes from the recorded metric contract and prior readings. Merge-time phases use the engine's checkpoint-resume.

Failure Modes Prevented

4 further anti-patterns (Solo-org adoption, sub-Step-2 adoption, AI write authority on the Insight Ledger, CEO/executive Compiler bypass) are Layer C anti-patterns owned by _common/GROWTH_BRAND_PROOF.md § Anti-Patterns Specific to Layer C — not re-derived here.

References

• _common/PROOF_CARRYING.md v3 — Tier A Foundation + Tier B Production Pipeline + cross-cutting G11/G14/G15
• _common/GROWTH_BRAND_PROOF.md — Layer C protocol (Market + Research + Brand axes), Insight Ledger, Incrementality Gate, Brand Compiler 3-layer, Growth-Brand Contract, Phased Adoption Step 1-4
• nexus/reference/acceptance-recipe.md — Phase 1 (merge-time) delegated chain
• nexus/reference/apex-recipe.md — discovery → ship cycle; Phase 6 can chain into growth-acceptance
• nexus/reference/summit-recipe.md — strategic decisions; can author Layer C content for growth-acceptance gating

Layer C's per-axis skill roster (Research / Brand / Market / Auto-action) is not re-listed here — every skill already appears against its owning phase in § Phase Contract above.