Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Analyze unit economics to evaluate per-unit profitability and business model scalability. Use this skill when the user needs to assess whether each transaction, customer, or product unit is profitable, evaluate startup viability, or optimize contribution margins — even if they say 'does our business model work', 'what's our margin per order', or 'can we scale profitably'.
.claude/skills/asgard-ai-platform-biz-unit-economics/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 46% | 0% |
Unit economics measures profitability at the single-unit level — per customer, per transaction, or per product. If unit economics are negative, scaling makes losses worse, not better. Positive unit economics are the prerequisite for sustainable growth.
IRON LAW: If Unit Economics Are Negative, Growth = Faster Death
Scaling a business with negative unit economics means losing more money
faster. "We'll make it up in volume" only works if variable costs per unit
decrease with scale (economies of scale). If they don't, more volume = more losses.
Prove unit economics are positive BEFORE investing in growth.Subscription/SaaS: | Metric | Formula | |--------|---------| | CAC | Total acquisition spend / New customers | | LTV | ARPU × Gross Margin × (1/Churn Rate) | | LTV:CAC | Must be > 3:1 | | Payback Period | CAC / (Monthly ARPU × Gross Margin) | | Net Revenue Retention | (Starting MRR + Expansion - Contraction - Churn) / Starting MRR |
E-commerce/Marketplace: | Metric | Formula | |--------|---------| | AOV | Total Revenue / Number of Orders | | Contribution Margin per Order | AOV - COGS - Shipping - Payment fees - Returns | | Repeat Rate | Returning customers / Total customers | | Fully-loaded CAC | Total marketing / New customers |
On-demand/Delivery: | Metric | Formula | |--------|---------| | Revenue per Trip | Order value × Take rate | | Cost per Trip | Driver pay + Support + Insurance + Incentives | | Contribution per Trip | Revenue - Cost per Trip | | Orders per Day per Driver | Utilization metric |
markdown# Unit Economics: {Business} ## Unit Definition - Unit: {customer / order / trip} - Time period: {monthly / per-transaction} ## Per-Unit Economics | Metric | Value | |--------|-------| | Revenue per unit | ${X} | | Variable cost per unit | ${X} | | Contribution margin | ${X} ({X%}) | ## Scalability Assessment | Volume | Unit Revenue | Unit Cost | CM | Notes | |--------|-------------|-----------|-----|-------| | Current ({N}) | ${X} | ${X} | {X%} | baseline | | 3x ({N}) | ${X} | ${X} | {X%} | {economies of scale?} | | 10x ({N}) | ${X} | ${X} | {X%} | {what changes?} | ## Verdict {Positive/negative unit economics, scalability assessment, key lever to improve}
Scenario: Unit economics for a Taiwan food delivery startup
| Script | Description | Usage | |--------|-------------|-------| | scripts/unit_economics.py | Compute SaaS unit economics (NRR, GRR, LTV, CAC payback, magic number, burn multiple) | python scripts/unit_economics.py --help |
Run python scripts/unit_economics.py --verify to execute built-in sanity tests.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-22 | pass→pass | 10,982 | 8,888 | -19% | 1 | 1 | 0% | 1,523 | 2,849 | +87% | 0 | 0 | — |
case-01 | fail→pass | 33,090 | 20,103 | -39% | 1 | 1 | 0% | 3,457 | 2,884 | -17% | 0 | 0 | — |
case-02 | fail→fail | 29,463 | 17,891 | -39% | 1 | 1 | 0% | 4,919 | 4,237 | -14% | 0 | 0 | — |
case-03 | fail→pass | 20,279 | 20,472 | +1% | 1 | 1 | 0% | 3,076 | 4,644 | +51% | 0 | 0 | — |
case-04 | pass→pass | 3,693 | 6,882 | +86% | 1 | 1 | 0% | 794 | 2,392 | +201% | 0 | 0 | — |
case-05 | pass→fail | 29,853 | 34,539 | +16% | 1 | 1 | 0% | 4,869 | 7,109 | +46% | 0 | 0 | — |
case-06 | pass→pass | 8,722 | 14,030 | +61% | 1 | 1 | 0% | 1,973 | 4,520 | +129% | 0 | 0 | — |
case-07 | pass→pass | 7,436 | 10,186 | +37% | 1 | 1 | 0% | 1,712 | 3,331 | +95% | 0 | 0 | — |
case-08 | pass→pass | 6,359 | 7,246 | +14% | 1 | 1 | 0% | 1,011 | 2,490 | +146% | 0 | 0 | — |
case-09 | pass→pass | 7,305 | 6,359 | -13% | 1 | 1 | 0% | 1,128 | 2,426 | +115% | 0 | 0 | — |
case-10 | pass→pass | 14,686 | 17,478 | +19% | 1 | 1 | 0% | 2,613 | 4,118 | +58% | 0 | 0 | — |
case-16 | pass→pass | 15,417 | 14,654 | -5% | 1 | 1 | 0% | 2,401 | 3,669 | +53% | 0 | 0 | — |
case-11 | pass→pass | 12,032 | 14,164 | +18% | 1 | 1 | 0% | 2,030 | 3,616 | +78% | 0 | 0 | — |
case-12 | pass→pass | 8,430 | 15,500 | +84% | 1 | 1 | 0% | 1,457 | 3,726 | +156% | 0 | 0 | — |
case-13 | fail→pass | 6,443 | 2,998 | -53% | 1 | 1 | 0% | 1,049 | 1,600 | +53% | 0 | 0 | — |
case-14 | pass→pass | 3,746 | 4,278 | +14% | 1 | 1 | 0% | 791 | 1,944 | +146% | 0 | 0 | — |
case-15 | fail→pass | 6,068 | 1,676 | -72% | 1 | 1 | 0% | 972 | 1,507 | +55% | 0 | 0 | — |
case-17 | pass→pass | 19,416 | 24,541 | +26% | 1 | 1 | 0% | 2,821 | 4,349 | +54% | 0 | 0 | — |
case-18 | pass→pass | 4,389 | 8,742 | +99% | 1 | 1 | 0% | 877 | 2,911 | +232% | 0 | 0 | — |
case-19 | pass→pass | 3,744 | 10,181 | +172% | 1 | 1 | 0% | 697 | 2,989 | +329% | 0 | 0 | — |
case-20 | fail→fail | 18,217 | 21,240 | +17% | 1 | 1 | 0% | 3,450 | 5,475 | +59% | 0 | 0 | — |
case-21 | pass→pass | 15,563 | 10,369 | -33% | 1 | 1 | 0% | 2,273 | 2,604 | +15% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.