| name | unit-economics-validation |
| version | 0.1.0 |
| status | PROBATIONARY |
| description | CFO-side gate for CMO ad-spend requests. Validates a proposed campaign on CAC, LTV, contribution margin, payback period, and target close rate BEFORE Atlas approves any portion of the cap. Approves, partial-approves, or rejects with a one-paragraph rationale. |
| metadata | {"category":"spend-governance","owner":"atlas","consumers":["maven-spend-requests","bravo-runway-decisions"],"tier":"full"} |
| triggers | ["validate-unit-economics","check-ltv-cac","ad-spend-gate","spend-approval","maven-request","review-campaign-math"] |
| sources | ["Benjamin Graham — margin of safety (unit economics must clear with buffer, not at parity)","Warren Buffett / Berkshire letters — owner earnings, durable advantage, reinvestment ROI","David Skok / For Entrepreneurs — SaaS LTV:CAC > 3, CAC payback < 12 months","Alex Hormozi — Value Equation, CFO-side pairs with Maven's offer-side","Philip Fisher — 15 points (retention, operator skill, real growth vs vanity)","CPA Canada — audit-defensible cost classification (T2125 line 8521)"] |
unit-economics-validation — CFO Spend Gate
Maven draws up the campaign. Atlas checks the math. No ad dollar leaves until the unit economics clear.
When to invoke
Automatic whenever:
data/pulse/cmo_pulse.json.current_spend_request_cad > 0- A new brand or channel is proposed
- CAC or close rate moves ±20% from the last approved baseline
- Spend request > 25% of current
approved_ad_spend_monthly_cap_cad
Manual: any time the operator or Bravo asks "should we greenlight this spend?"
Required inputs (Maven must provide; Atlas rejects if missing)
| Field | Definition | Source of truth |
|---|
channel | Meta, Google, TikTok, cold email, organic | Maven campaign brief |
brand | OASIS / Bennett-shared / SunBiz / PropFlow / Nostalgic | Maven pulse |
proposed_monthly_spend_cad | Total budget before buffer | Maven request |
target_cpl_cad | Cost per lead, forecasted | Maven forecast / historical |
target_close_rate_pct | Lead → paying customer conversion | Bravo CRM / historical |
avg_deal_size_cad | First-touch revenue | Stripe / invoice history |
gross_margin_pct | After cost of delivery (excl. ad spend) | CFO books / brand_economics in cfo_pulse |
expected_retention_months | Median customer lifetime | Cohort data if available, else sector benchmark |
time_to_payback_months | Months until cumulative GM >= CAC | Derived |
If any field is forecast-only (no historical evidence), flag it SPECULATIVE and demote the approval to experimentation-only.
The 5 Checks
Check 1 — LTV:CAC ratio (Skok's 3:1 rule)
CAC = proposed_monthly_spend_cad / (leads × close_rate)
LTV = avg_deal_size_cad × gross_margin_pct × expected_retention_months
(for subscription products: use MRR × gross_margin × avg_lifetime_months)
LTV:CAC ratio = LTV / CAC
PASS if ratio >= 3.0
WATCH if 2.0 <= ratio < 3.0 (allow only if at least one lever has headroom)
FAIL if ratio < 2.0
Rationale: Skok's canonical benchmark. Below 3:1 the business can grow but generates no surplus to compound. Below 2:1 growth actively destroys capital.
Check 2 — CAC payback period (the cashflow check)
monthly_gross_margin_per_customer = avg_deal_size_cad × gross_margin_pct
/ expected_retention_months
payback_months = CAC / monthly_gross_margin_per_customer
PASS if payback_months <= 12 (SaaS default)
PASS if payback_months <= 6 (services / one-time revenue)
WATCH if 12 < payback_months <= 18
FAIL if payback_months > 18
Rationale: Atlas guards the capital floor. Long payback periods compound into runway problems if retention misses. Canadian sole-prop tax treatment amplifies the hit (spend is deductible but cash is cash).
Check 3 — Positive contribution margin at target close rate
contribution_per_customer = (avg_deal_size_cad × gross_margin_pct) - CAC
PASS if contribution_per_customer > 0 AND close_rate IS historical (not forecast)
WATCH if contribution_per_customer > 0 but close_rate IS a forecast (allow 25% of cap)
FAIL if contribution_per_customer <= 0
Rationale: Graham's margin of safety. Positive contribution with forecast close rate = theoretical; with historical close rate = validated.
Check 4 — Concentration and runway drag
Read cfo_pulse.json.liquid_cad and montreal_floor_target_cad.
Read ceo_pulse concentration_risk_single_client_pct.
IF liquid_cad < montreal_floor_target_cad:
cap_multiplier = 0.5
IF concentration_risk_single_client_pct >= 70:
cap_multiplier *= 0.5
IF concentration_risk_single_client_pct >= 80:
spend_veto = True
approved_cap_cad = approved_ad_spend_monthly_cap_cad × cap_multiplier
Rationale: Capital preservation beats growth. No campaign is worth taking the operator below the relocation floor. No campaign is worth doubling down on a concentrated revenue book.
Check 5 — T2125 deductibility
Spend on line 8521 (advertising) is a dollar-for-dollar deduction against self-employment income. Post-CCPC it deducts at the CCPC bracket (~12.2% in Ontario on first $500K). Atlas logs the expected tax offset so the after-tax spend number is honest.
after_tax_spend = proposed_monthly_spend_cad × (1 - marginal_rate)
Note: marginal_rate is a savings, NOT a justification. A deductible bad spend is still a bad spend.
Decision matrix
| Check 1 | Check 2 | Check 3 | Check 4 | Decision |
|---|
| PASS | PASS | PASS | OK | APPROVE up to cap |
| PASS | PASS | WATCH | OK | PARTIAL at 50% of cap, require 14-day CAC feedback |
| PASS | WATCH | PASS | OK | PARTIAL at 50% of cap, require payback re-proof by day 30 |
| FAIL anywhere | — | — | — | REJECT with specific failing metric + what would unblock |
| any | any | any | veto | REJECT regardless — concentration/floor breach |
Output format Atlas writes into cfo_pulse.json
"maven_spend_decision": {
"request_id": "<uuid or timestamp>",
"campaign": "pulse-lead-gen",
"brand": "OASIS",
"requested_cad": 1200,
"approved_cad": 100,
"decision": "PARTIAL",
"checks": {
"ltv_cac": { "value": 4.2, "verdict": "PASS" },
"payback_months": { "value": 9.1, "verdict": "PASS" },
"contribution_margin": { "value": 180, "verdict": "WATCH", "reason": "close_rate is forecast" },
"concentration": { "value": 94, "verdict": "VETO_ACTIVE", "cap_multiplier": 0.25 },
"deductibility": { "line": "T2125 8521", "marginal_rate": 0.2565 }
},
"rationale": "LTV:CAC healthy, payback inside 12 months. Close rate is a forecast not historical, so downgraded to experimentation only. Bennett concentration at 94% caps total monthly spend at $100 CAD until diversification below 70%.",
"unblock_at": ["Bennett concentration < 70%", "14-day historical close rate ≥ forecast", "liquid_cad ≥ $10K floor"]
}
Anti-patterns Atlas actively rejects
- "Trust the creative" — creative variance is real but can't rescue broken unit economics. Reject.
- "Scale reveals retention" — retention gets WORSE at scale (light buyers have lower loyalty per Sharp). Don't budget based on early-cohort retention.
- "We'll make it back on upsell" — model upsell separately with its own LTV. Primary purchase must stand on its own.
- "Competitors spend more so we should" — competitor spend is not a signal of their unit economics. They may be burning capital too.
- "Lifetime value is theoretical anyway" — no: without LTV, every campaign is a gamble. Require historical cohort data or use conservative sector median.
When to probationary-promote → VALIDATED
After 3 successful decisions where:
- The approval rationale held (6-month post-approval review)
- Maven hit CAC within ±15% of target
- No capital-floor breach occurred
Log review outcomes in memory/PATTERNS.md as they happen.
Cross-references
- [[CFO_CANON]] — the investors/frameworks this skill derives from (pillars 1–5 especially)
- [[AGENT_ORCHESTRATION]] — pulse protocol, spend handshake sequence
- [[skills/financial-health-check/SKILL|financial-health-check]] — concentration and runway inputs
- [[skills/position-sizing/SKILL|position-sizing]] — parallel sizing logic for equity positions
- [[skills/self-improvement-protocol/SKILL|self-improvement-protocol]] — the loop that promoted this skill from gap to build
- [[STATE]] — live
liquid_cad + approved_ad_spend_monthly_cap_cad
cfo_pulse.json.brand_economics — per-brand gross margin inputs to Check 1- [[INDEX]] — graph hub
Success metrics for this skill itself
| Metric | Target | Review cadence |
|---|
| Approval accuracy | ≥ 80% of approvals hit target CAC within ±15% | Monthly |
| Rejection accuracy | ≥ 80% of rejections would have burned capital if approved | Quarterly (retrospective on rejected campaigns) |
| Time-to-decision | < 10 minutes from Maven request → pulse write | Per-request |
| CC override rate | < 10% of decisions overridden by operator | Monthly |
If any metric trips below target for two consecutive reviews, trigger self-improvement Protocol 4 Reflexion on this skill.
