Startup Business Models and Pricing

Epstein outlines nine common models: SaaS, transactional, marketplaces, hard tech, usage-based, enterprise, advertising, e-commerce, and biotech. In YC’s top 100 companies, SaaS accounts for 31, transactional businesses 22, and marketplaces 14, together making up 67 companies.

Half the value of YC’s top 100 companies comes from the top ten, where five are marketplaces. Marketplaces make up 14 companies but 30% of the list’s total value, while transactional businesses make up 22 companies and 29% of its value. Epstein connects these outcomes to marketplace network effects and to transactional companies’ position in the flow of funds.

Epstein says services, affiliate businesses, hardware, and businesses dependent on another platform have scaling, margin, capital, or platform risks. Recurring revenue can make growth more predictable, but it depends on retention: at 95% monthly retention, 100 customers would fall to 54 after a year; at 90%, they would fall to 28. He also describes moats such as network effects, switching costs, technical innovation, scale economies, and organic distribution.

Epstein recommends charging early because customers’ willingness to pay can reveal which segment values the product and how much value they see. He says to price around customer value rather than cost, and to raise prices gradually to learn where customers push back. Stripe tested perceived value by initially charging 5% per transaction when competitors charged about 3%; Segment raised its annual price from $120 to $18,000 after testing a higher enterprise price.

Epstein says pricing can change as a product improves: companies can grandfather existing customers or give notice before raising their prices. He contrasts Quicken’s complex pricing page with GitLab’s three clear plans, arguing that complexity can create friction. He closes with five recommendations: charge, price on value, expect to be undercharging, iterate, and keep pricing simple.