Definition (canonical)
A business model is the logic by which a company creates value for customers and captures part of it for itself — who pays, for what, how much, how often, and at what cost of delivery. Products compete on features; companies compete on models — and model choices routinely beat product quality in deciding who wins.
The anatomy in five questions
Strip any canvas to its skeleton: Who is the customer (and who is the payer — advertising and marketplaces split the two)? What problem is solved and how is that valuable? How is money taken — one-off sale, subscription, commission, usage, licensing, freemium's paid slice? What does delivery cost — and do unit economics improve or worsen with scale? What defends the position — switching costs, network effects, brand, data, regulation? Weak answers to question five explain most "great product, dead company" stories.
The major families, compared by physics
| Model | Revenue physics | Scale behavior |
|---|---|---|
| Direct sales (product/service) | linear: sell again each period | margins stable or eroding |
| Subscription | compounding recurring layers | retention decides everything |
| Marketplace / platform | take rate on others' transactions | winner-take-most via network effects |
| Advertising | attention resold to advertisers | needs massive scale to matter |
| Licensing / IP | near-zero marginal cost | slow to build, lucrative to hold |
Hybrids dominate real companies (hardware + subscription; free tool + marketplace) — the design question is which engine is PRIMARY, because that engine's physics dictates metrics, funding needs and org design.
Stress-testing before scaling
Four tests catch most fragile models: the unit test — is one more customer profitable on honest math? the concentration test — does survival hang on one channel, supplier or client? the incentive test — does your revenue grow when the customer wins, or when they lose (overage fees, breakage)? Misaligned models get regulated or disrupted. the copy test — if a well-funded competitor clones the product tomorrow, what stops customers leaving? Models flunking two or more tests need redesign, not more marketing.
FAQ
Business model vs strategy vs plan? The model is the earning machine's design; strategy is where and how to compete; the plan is numbers over time. Confusing them produces beautiful plans for broken machines.
Can a company change models? Pivots of capture (license → SaaS) succeed regularly; pivots of customer are near-restarts. Price the difference honestly.
Where do most new models fail? Value capture — creating value customers happily take for free is common; charging for it is the model.
Related: Subscription Model · Marketplace · Unit Economics · Network Effects Reviewed: 2026-07-15 · Sources: business model canon