The most expensive mistake in startup history is being wrong about unit economics. Kozmo.com (raised more than a quarter-billion dollars in 1999, delivering ice cream and DVDs in under an hour with no minimums or fees) is the textbook case. Magazine covers, millions of customers, and a margin that lost money on every order. They confused traction for viability.
The trap is still the most common one we see. We have run the math on our own ventures and on partner-venture due diligence. The framework below is what we apply before any growth conversation.
Key Takeaway
A sustainable business is one where the customer can be acquired profitably, served at positive margin, and kept long enough to recover acquisition cost with money left over. Four numbers — CAC, LTV, Payback Period, Contribution Margin — settle whether the business model is physics-positive. Everything else on the dashboard is a derivative of these four.
The Problem
Founders confuse revenue, gross profit, and contribution margin. They use blended CAC when paid CAC is the truth-teller. They project LTV from optimistic retention assumptions instead of measuring it from actual cohort behavior. The mistake is rarely "we did the math wrong." The mistake is "we did the wrong math."
The Framework
01 — CAC: The Hidden Complexity
What we look for:
- Fully-loaded CAC including sales salaries and tooling, not just paid spend
- Paid CAC measured separately from blended CAC — the gap is the organic advantage
- CAC measured per channel, per segment, per cohort — never as one number
Why it matters: A blended CAC hides which channel is healthy and which is bleeding. If word-of-mouth contributes a meaningful share of customers, the blended number flatters the underlying paid economics. The marginal next customer is acquired at paid CAC; the blended figure is a story about the average past, not the next dollar of growth.
02 — LTV: Where Optimism Goes to Die
What we look for:
- LTV calculated from actual monthly churn, not projected retention
- Gross-margin-adjusted, not revenue-only — LTV at 30% gross margin is fundamentally different from LTV at 80%
- Expansion revenue measured separately, not assumed
Why it matters:
The formula (Monthly Revenue × Gross Margin) ÷ Monthly Churn forces honesty. The "average lifespan" formulation invites optimism — eight months of operating data does not justify an "indefinite lifespan" assumption. Most founder LTV estimates are wrong by a factor that the cohort data will eventually settle.
03 — Payback Period: The Real Cash Constraint
What we look for:
- Payback period measured in months, not justified in narrative
- Less than one-third of expected customer lifespan
- Sensitive to gross margin — a margin compression compounds straight into payback
Why it matters: LTV:CAC tells you if the economics work; payback period tells you how fast. Two businesses with the same 3:1 ratio can have a six-month payback or a two-year payback — they need radically different working capital to grow at the same pace. Cash flow lives and dies on the payback math, not on the ratio.
04 — Contribution Margin: The Forgotten Number
What we look for:
- Revenue minus all variable costs, including delivery, support, and per-transaction fees
- Positive at the unit level — and increasingly so as scale absorbs fixed costs
- Modeled at the per-transaction grain, not just at the company P&L
Why it matters: A negative contribution margin means every additional customer loses you money — scale accelerates the loss, not the recovery. Kozmo and MoviePass are both case studies of this exact failure mode. The discipline is brutal but simple: if the unit economics are negative at the transaction level, no amount of growth fixes them.
Implementation Checklist
- Calculate paid and blended CAC separately — by channel, not blended
- Pull a 12-month cohort and compute LTV from actual data, not projections
- Compute payback period in months and compare to expected lifespan
- Verify contribution margin is positive at the transaction level
- Identify the one channel with the worst paid CAC and pressure-test the spend
What This Produces
- Honest visibility into which growth is profitable and which is loss-leading
- A working-capital plan that matches the actual cash-conversion cycle
- A kill criterion for channels that look efficient on the blended dashboard
Common Mistakes
- Mixing blended and paid CAC. Blended is a story about the past; paid is the truth about the next dollar.
- LTV from optimistic projections. Use the cohort data you have. The "we expect retention to improve" claim should not enter the model until it has happened.
- Treating contribution margin as an accounting question. It is the survival question. A negative number means scale accelerates the problem.
Next Steps
If you are running unit-economic discipline on a venture you operate, our free training on execution systems walks the four numbers with worked examples. To see the math applied across our portfolio, explore the ventures.
Arena-forged across 8 venture lines. Every framework tested on our own operations before it reaches a partner. See Bayanihan Harvest for the proof.