Startup Business Models and Pricing | Startup School
Choosing a business model can feel like a creative exercise, but Y Combinator group partner Aaron Epstein argues that founders rarely need to invent a new commercial structure. The strongest startups often introduce a novel product while relying on an established way of making money.
In this Startup School session, Epstein examines the models behind billion-dollar companies, patterns among Y Combinator’s top 100 businesses, and the pricing lessons founders can extract from them. His central message is practical: use a model with proven economics, build genuine defensibility, and treat pricing as an ongoing source of customer insight.
Nine Business Models Behind Billion-Dollar Companies
Epstein identifies nine broad models that account for most billion-dollar technology companies. These categories are not rigid—a company may combine several—but they give founders a useful framework for understanding how revenue, growth, and margins fit together.
Software as a service (SaaS) companies sell ongoing access to software, usually through monthly or annual subscriptions. Recurring payments can make revenue more predictable, while software delivery allows the company to serve additional customers without expanding its workforce at the same rate.
Transactional businesses earn money whenever value moves through their systems. Payment processors and financial platforms are common examples. Because revenue is linked directly to customer activity, these companies can grow as their customers transact more.
Marketplaces connect two or more groups, such as buyers and sellers, and collect a fee for enabling the exchange. Their appeal lies partly in network effects: each additional participant can make the marketplace more useful for everyone else.
Hard-tech companies commercialize difficult engineering or scientific advances. Their products may require substantial capital and long development cycles, but technical complexity can also create a powerful barrier to entry.
Usage-based businesses charge according to consumption. Customers might pay for computing resources, messages sent, data processed, or another measurable unit. Revenue can expand naturally as a product becomes more deeply embedded in a customer’s operations.
Enterprise companies sell high-value products to large organizations. These deals often involve sales teams, procurement processes, security reviews, and long contracts. Although sales cycles can be slow, successful vendors can earn substantial revenue from each account.
Advertising businesses provide content or services to users and monetize their attention. Epstein cautions that this model generally requires enormous reach because the revenue earned from an individual user is often limited.
E-commerce companies sell physical goods online. They may achieve significant scale, but inventory, fulfillment, returns, and competition can put pressure on margins.
Biotechnology companies turn advances in biology and medicine into valuable treatments, diagnostics, or platforms. Like hard tech, biotech usually carries unusual technical, regulatory, and capital requirements.
Epstein’s point is not that founders must fit perfectly inside one box. Rather, understanding these models helps them avoid building a product whose economics depend on assumptions that have rarely worked at scale.
What YC’s Top 100 Companies Reveal
Epstein draws several conclusions from Y Combinator’s top 100 companies. Most strikingly, 67% use one of three models: SaaS, transactional, or marketplace. That concentration suggests that certain economic structures repeatedly produce large outcomes.
The value within the group is also highly uneven. According to Epstein, half of the combined value comes from just ten companies. Startup results follow a power-law pattern: a small number of exceptional winners account for a disproportionate share of the total.
Marketplaces are especially capable of producing such outliers. Once a marketplace establishes liquidity and attracts participants on both sides, competitors face a difficult challenge. Buyers prefer the platform with the strongest supply, while sellers go where the buyers already are. That feedback loop can produce a winner-take-most market.
Transactional companies benefit from occupying a strategic position in the flow of money. Instead of charging separately for a supporting tool, they participate whenever an economic event occurs. As transaction volume rises, revenue can rise alongside it.
SaaS companies gain a different advantage from recurring subscriptions. Their income is generally easier to forecast than one-time sales, and retained customers continue contributing revenue without needing to be reacquired every month. That does not make SaaS automatically successful, however. If customers regularly cancel, the apparent stability of recurring revenue quickly disappears.
Advertising and e-commerce are less prominent among the top 100 companies Epstein analyzes. Advertising, in particular, tends to reward only the largest destinations. His assessment is blunt: a startup pursuing this model may need to become one of the internet’s most heavily visited properties before the economics become truly exceptional.
Why Services Rarely Produce Venture-Scale Outcomes
No services or consulting businesses appear in Epstein’s top-100 analysis. He attributes that absence to the way service revenue scales. A consultancy generally needs more people to complete more projects, which makes growth dependent on hiring and management rather than software distribution.
Project-based income is also less predictable than recurring revenue. Even after completing excellent work, a service company must keep winning new engagements. Labor costs can remain high, and margins tend to be lower than those of software products serving many customers from a shared platform.
This does not mean consulting is a bad business. It can be profitable, durable, and valuable to clients. Epstein’s narrower argument concerns the kind of rapid, high-margin scaling sought by venture-backed technology companies. For that goal, software usually offers more leverage than labor.
The Characteristics of a Durable Startup
Across the business models, Epstein highlights several traits that frequently appear in the strongest companies.
Recurring revenue is one of the most important. It improves visibility into future performance and can reduce the burden of repeatedly acquiring the same amount of revenue. But recurrence only matters when paired with retention. A subscription product that customers abandon quickly has merely spread a weak sale across several payments.
Defensibility matters just as much. Epstein points to moats such as:
Network effects that increase a product’s value as more people participate
Customer lock-in or meaningful switching costs
Technical breakthroughs that competitors cannot easily reproduce
Superior margins and unit economics
A strategic position close to a transaction or essential workflow
Scalability is another recurring theme. Strong technology businesses can serve far more customers without increasing headcount proportionally. Software, automation, and self-service distribution allow revenue to separate from labor growth.
Together, these observations produce a clear operating principle: founders can concentrate their originality in the product while borrowing a commercial model that has already demonstrated its ability to scale. Innovation does not require inventing both the solution and the underlying economics from scratch.
Pricing as a Customer-Discovery Tool
Epstein presents pricing as more than a financial decision. It is an experiment that reveals who values a product, how urgently they need it, and which customer-acquisition channels can support the resulting economics.
The first lesson is simple: charge something. Founders often postpone monetization because they fear losing users or pushing them toward a competitor. Yet free adoption does not prove that customers receive enough value to sustain a business. Payment creates a stronger signal.
Different prices also expose differences among customer segments. A casual user and an enterprise with a costly operational problem may use the same product but value it very differently. Raising the price can help founders discover which group has the greatest willingness to pay.
Epstein warns against spending too long searching for a mathematically perfect launch price. Early-stage companies generally need the correct order of magnitude, not a permanent answer. Pricing can change as the product improves, the target customer becomes clearer, and the company learns how buyers respond.
Price for Value, Not Just Cost
Cost-plus pricing begins with the expense of serving a customer and adds a margin. Although costs cannot be ignored, Epstein argues that they should not determine the price by themselves. A product’s economic value to the customer may be far greater than its delivery cost.
Founders can test that value by gradually increasing prices until buyers begin resisting. The goal is not to make every prospect happy. It is to identify the point at which price, demand, and customer value support a healthy company.
Competing primarily on low price is rarely defensible. Another startup can usually discount too, and a bargain price may even weaken the product’s perceived credibility. Higher prices can fund better service, stronger product development, and more expensive acquisition channels. They may also signal that the product is built to solve an important problem.
For an established product, increasing prices can be one of the most direct ways to grow revenue. If customers refuse to pay more, Epstein suggests examining the underlying value proposition. The company may need to make the product more useful or address a larger, more urgent problem.
Founders can still negotiate selectively. A customer might receive a lower rate in exchange for a longer commitment, a case study, faster payment, or another benefit. The important distinction is that the discount buys something useful rather than simply reflecting uncertainty.
Pricing should also remain easy to understand. Excessive tiers, variables, and exceptions add friction to a purchase. A simple model helps customers evaluate the offer and helps the startup interpret the results of its experiments.
How Segment Discovered It Was Underpricing
Epstein closes with the story of Segment, a customer-data infrastructure company that initially offered its product for free. When the team needed revenue growth to strengthen its fundraising position, it introduced a price of $10 per month, or $120 per year.
The response challenged the founders’ assumptions. Some customers did not object that the price was too high; instead, they worried it was too low. Because they depended on Segment, they wanted confidence that the company could remain reliable and continue operating. The tiny fee suggested that the business might not be sustainable.
A sales adviser later encouraged Segment to view its product as enterprise infrastructure and proposed an annual price of $120,000. The founders were uncomfortable presenting such a dramatic increase, and the customer did not accept that figure. But the negotiation eventually produced an $18,000 annual contract—150 times the original yearly price.
That result did not prove that every customer would pay the same amount. It showed that Segment had misunderstood the scale of the problem it was solving for serious business users. The experience helped the company pursue progressively larger deals and develop a commercial foundation for its growth. Segment was eventually acquired by Twilio for more than $3 billion.
The broader lesson is not to raise prices recklessly. It is to test assumptions directly. Customers may perceive more value than founders expect, and an implausibly low price can obscure that value rather than unlock demand.
