Purpose
Turn a program's revenue and cost inputs into a rigorous per-unit P&L that exposes contribution margin, the ranked cost drivers eroding it, and the capital efficiency of scaling — so a PM or GM knows whether the unit is profitable, at what volume it breaks even, and which one or two levers change the answer.
When to Use
Invoke this skill when the user asks to:
- Determine whether a marketplace transaction, take rate, or GMV cohort is profitable per unit
- Analyze the unit economics of a fintech product (a loan, a card, a payment flow, an account)
- Break down contribution margin and cost stack for a subscription tier or platform program
- Diagnose why a growing product is burning cash despite rising revenue
- Model CAC payback, LTV/CAC, and contribution margin for a specific customer or transaction segment
- Pressure-test a pricing or take-rate change before proposing it
Inputs Required
Before starting, collect:
- Business model: Marketplace / fintech / SaaS-platform / hybrid — and how it makes money
- Unit definition: What "one unit" is (a transaction, an active user, a loan, a subscription, a completed order)
- Revenue per unit: Take rate, ARPU, interest/interchange, subscription price, or gross fee per unit
- Direct cost components: Every variable cost tied to the unit (COGS, payment processing, fraud/loss, fulfillment, support, cloud/infra, incentives/promos)
- Acquisition cost: Blended or channel CAC, plus any onboarding or activation cost per unit
- Volume and retention: Units per period, repeat/retention rate or churn, expected lifetime
- Capital committed: Any float, inventory, credit balance, or working capital tied up per unit (fintech/marketplace especially)
If retention or lifetime is unknown, model a base case at a stated assumption and flag it as the highest-leverage unknown.
Instructions
Step 1: Define the Unit and the Time Horizon
Lock the denominator before any math. State explicitly:
- What one unit is (transaction vs. user vs. account vs. cohort) — mixing these is the most common error
- Per-transaction economics vs. per-user lifetime economics — decide which the question requires, or run both
- The time horizon (single transaction, monthly, annual, or full lifetime)
Restate the unit back to the user in one sentence so the rest of the analysis is unambiguous.
Step 2: Build the Per-Unit P&L (Contribution Margin Waterfall)
Construct a top-to-bottom waterfall from gross revenue to contribution margin:
- Gross revenue per unit: Take rate × GMV, ARPU, gross fee, or price
- Less: direct variable costs, itemized — payment processing, COGS/fulfillment, fraud and loss provision, variable support, variable infra, incentives/promos
- = Contribution margin (CM1) — before acquisition
- Less: acquisition/onboarding cost allocated per unit
- = Contribution margin after CAC (CM2)
Show each line as both an absolute figure and as a percentage of revenue. Name the largest two cost lines as the primary margin drivers.
Step 3: Cost Stack Diagnosis
Rank cost components by their share of revenue and separate them by behavior:
- Truly variable (scales linearly with volume): processing, loss, incentives
- Semi-variable (steps up in blocks): support, infra
- Structurally fixed but per-unit-allocated: onboarding, platform overhead Flag any cost line above 15% of revenue as a lever candidate, and note which costs shrink with scale versus which do not.
Step 4: Capital Efficiency and Payback
Assess how much cash the unit ties up and how fast it returns:
- CAC payback period: Months of contribution margin to recover acquisition cost
- LTV: Contribution margin per period × expected lifetime (state the retention assumption)
- LTV / CAC ratio: Flag below 3x as fragile, above 5x as under-invested in growth
- Working capital / float per unit: Cash committed before the unit turns profitable (float, inventory, credit balance) — critical for fintech and marketplace models
- Contribution-margin-positive point: The unit count or time at which cumulative CM covers fixed cost allocation
Step 5: Sensitivity Analysis
Build a table showing how contribution margin after CAC responds to movement in the top three drivers (typically take rate/price, the largest variable cost, and retention). Show each driver at −20%, base, and +20%, with the resulting CM2 and LTV/CAC. Identify which single variable the outcome is most sensitive to.
Step 6: Ranked Levers and Recommendation
Produce a prioritized list of 4–6 levers to improve unit economics, ranked by expected margin impact and feasibility. For each: the specific action, the line it moves, the estimated point-change in CM2 or payback, and the primary risk. Close with one clear verdict: is this unit economically sound as-is, sound only at scale, or structurally unprofitable — and the single most important thing to fix.
Output Format
- Structure: Unit definition → per-unit P&L waterfall → cost stack ranking → capital efficiency metrics → sensitivity table → ranked levers → verdict
- Length: 900–1,300 words plus two tables (P&L waterfall and sensitivity table)
- Format: Structured document with two tables and a ranked lever list; concise prose between
- Sections: Unit Definition → Per-Unit P&L → Cost Stack Diagnosis → Capital Efficiency & Payback → Sensitivity Analysis → Ranked Levers → Verdict
- Tone: Operator-analyst — quantitative, direct, willing to call a unit unprofitable
Quality Checklist
Before delivering output, verify:
- [ ] The unit is defined in one unambiguous sentence before any calculation
- [ ] The P&L waterfall shows every cost line as both an absolute figure and a percent of revenue
- [ ] Both CM1 (pre-CAC) and CM2 (post-CAC) are reported distinctly
- [ ] CAC payback, LTV/CAC, and capital-tied-up-per-unit are all computed with stated assumptions
- [ ] The sensitivity table covers exactly the top three drivers at −20% / base / +20%
- [ ] The ranked levers are ordered by impact and each names the specific P&L line it moves
- [ ] A single verdict is stated — no hedging between "profitable" and "not"
Notes
- Per-transaction and per-user lifetime economics answer different questions; when a product loses money per transaction but wins per user (fintech, marketplaces), run both and say so explicitly.
- Fraud/loss provisioning and float are the most commonly omitted costs in fintech models — probe for them even when not volunteered.
- The most common self-deception in growing platforms is treating semi-variable costs as fixed; scale does not rescue a unit whose truly variable costs exceed revenue.
- When retention is the sensitivity winner, the real problem is usually a retention problem masquerading as a margin problem — name it.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills