The 9 Business Models Behind Billion-Dollar Startups — And Why Most Founders Undercharge

Founders get frustrated when investors pass and growth stalls. They blame the product, the market, the timing. Usually the real problem is simpler: they picked the wrong business model.

A business model is how you make money. It sounds obvious, but the difference between a model that scales and one that stalls is often the difference between a billion-dollar outcome and a quiet shutdown. An analysis of Y Combinator’s top 100 companies reveals that nearly all of them use one of nine business models — and the three models that produce the most value share a specific property worth understanding before you commit.

This article covers those 9 business models, what makes each one work, which models are traps, and the pricing insights that separate founders who capture value from founders who leave it on the table.

The 9 Business Models That Build Winners

Y Combinator Group Partner Aaron Epstein studied the top 100 YC companies and found that SaaS, transactional, and marketplace models alone account for 67% of the portfolio. The remaining companies split across usage-based, enterprise, advertising, e-commerce, subscription, and hardware-plus-software models. Here is what each one looks like.

1. SaaS (Software as a Service)

Recurring subscription for software delivered over the internet. Customers pay monthly or annually for access.

Why it works: High gross margins (typically 75-90%), predictable revenue, and strong retention mechanics. Once a customer integrates your software into their workflow, switching costs make them sticky. Revenue compounds as you add customers without proportionally adding costs. SaaS represents 31% of the top 100 YC companies — the single most common model — because recurring revenue makes the economics so favorable.

Examples: Stripe (payment infrastructure), Gusto (payroll), HubSpot (marketing automation).

Key metric: Monthly Recurring Revenue (MRR) and Net Revenue Retention (NRR). The best SaaS companies achieve NRR above 120%, meaning they grow revenue from existing customers even before adding new ones. Snowflake reported NRR of 164% at its IPO; GitLab exceeded 150%. A company with 120% NRR grows its existing revenue base 2.5x over five years without acquiring a single new customer.

2. Transactional

Take a percentage cut of each transaction processed through your platform. Revenue scales directly with transaction volume.

Why it works: Revenue grows automatically as your customers grow. If you process payments for 10,000 businesses and each business grows 20% year over year, your revenue grows 20% without acquiring a single new customer. Alignment between your success and your customers’ success is built into the model. Transactional businesses represent 22% of the top 100 YC companies but generate 29% of the overall portfolio value.

Examples: Stripe (2.9% + 30 cents per domestic transaction in the US), Brex, Square.

Key insight: Many of the most valuable fintech companies combine the SaaS model with transactional fees — a subscription for the software platform plus a percentage of every transaction processed through it. This hybrid captures both predictable base revenue and growth-linked upside.

3. Marketplace

Connect buyers and sellers. Take a percentage of each transaction that occurs through your platform. The platform creates value by aggregating supply and demand in one place.

Why it works: Network effects. As more sellers join, the marketplace becomes more valuable to buyers. As more buyers join, it becomes more valuable to sellers. This flywheel, once spinning, creates a moat that is extremely difficult for competitors to replicate. Marketplaces make up only 14% of the top 100 YC companies but generate 30% of the total portfolio value — the highest value concentration of any model. Five of the top ten YC companies by valuation are marketplaces.

Why it is hard: The chicken-and-egg problem. You need sellers to attract buyers and buyers to attract sellers. Getting the flywheel started requires creative strategies — many successful marketplaces started by manually recruiting one side before opening to the other.

Examples: Airbnb, DoorDash, Instacart, Faire (wholesale marketplace valued at $5.2 billion as of late 2025).

4. Usage-Based

Charge based on consumption — the more a customer uses, the more they pay. No fixed subscription; billing is metered.

Why it works: Revenue grows with customer success. If a customer’s usage doubles, your revenue from that customer doubles. This alignment makes it easy for customers to start small (low commitment) and scale up naturally. It also creates a powerful signal: if usage is growing, the product is delivering value.

Examples: AWS (compute hours, storage), Twilio (per API call), Snowflake (consumption-based compute credits). OpenAI charges per token processed — a pure usage-based model where costs scale directly with how much of the AI a customer consumes.

Key nuance: Usage-based pricing works best when the unit of consumption is clear, measurable, and correlated with the value the customer receives. If usage does not correlate with value, customers feel they are being charged unfairly.

5. Subscription (Non-SaaS)

Recurring payment for non-software products — physical goods, content, access, or services delivered on a regular schedule.

Why it works: Predictable revenue and customer retention, applied to categories beyond software. Converts one-time purchases into ongoing relationships.

Examples: Dollar Shave Club (razors — acquired by Unilever for $1 billion in 2016), Stitch Fix (clothing), Netflix (content).

Caveat: Physical subscription businesses face logistics and fulfillment costs that SaaS does not. Margins are typically lower, and churn can be higher because the switching cost is simply canceling a shipment. Dollar Shave Club’s story illustrates both the potential and the limits: the $1 billion acquisition looked like a triumph, but Unilever sold the company to private equity in 2023 after struggling to grow it further.

6. E-Commerce

Sell physical or digital goods directly to consumers. Revenue comes from the margin between cost of goods and selling price.

Why it works at scale: When combined with a strong brand or proprietary product, direct-to-consumer e-commerce can achieve healthy margins. The challenge is that without a differentiated product, you are competing on price — and competing on price against Amazon is not a winning strategy.

Examples: Warby Parker (eyewear), Allbirds (shoes), Glossier (beauty).

Reality check: The DTC wave that made these brands famous has hit turbulence. Rising customer acquisition costs, slowing e-commerce growth, and an increasingly crowded landscape have made profitability elusive. Allbirds has struggled significantly since its IPO, and Glossier went through layoffs and restructuring. The lesson: a DTC brand needs either truly differentiated products or a path to profitability that does not depend on cheap digital advertising.

7. Advertising

Offer a free product to users. Monetize by selling user attention — ad impressions, sponsored content, or promoted listings — to advertisers.

Why it works at scale: Zero marginal cost per user means you can grow to hundreds of millions of users without proportional cost increases. When it works, it creates some of the most valuable companies in the world.

Why it is dangerous for startups: Requires massive scale before revenue becomes meaningful. A startup with 10,000 users and an advertising model generates almost nothing. The same startup with a SaaS model and 10,000 paying users at $50/month generates $500,000 in monthly revenue. Only 3% of the top 100 YC companies use advertising as their primary model. Advertising is a business model for companies that have already achieved massive scale, not for companies trying to achieve it.

Examples: Google, Reddit, Meta. All of these reached hundreds of millions of users before advertising revenue became their primary engine.

8. Enterprise

Large contracts with big companies. Long sales cycles, dedicated account managers, custom implementations, multi-year agreements.

Why it works: Enormous contract values — a single enterprise deal can be worth millions per year. Once you are inside a large organization, switching costs and integration depth make it extremely difficult for competitors to displace you. Databricks, valued at $134 billion after its December 2025 funding round, grew revenue over 55% year-over-year to a $4.8 billion run rate, with more than 700 customers consuming over $1 million annually.

Why it is hard: Sales cycles of 6-18 months. Requires a dedicated sales team, solutions engineers, and customer success managers. The cost of acquiring each customer is high, and the time-to-revenue is long. Not a model for founders who need fast feedback loops.

Examples: Palantir, Snowflake (enterprise tier), Databricks.

9. Hardware + Subscription

Sell a physical device, then charge an ongoing subscription for the service that makes the device valuable.

Why it works: The hardware creates a tangible, physical relationship with the customer. The subscription generates recurring revenue and captures the ongoing value of the service. Together, they create stronger lock-in than either model alone.

Examples: Peloton (bike + class subscription — subscriptions now account for roughly two-thirds of its revenue), Ring (camera + monitoring subscription, acquired by Amazon for approximately $1 billion in 2018), Apple (devices + Apple One services bundle).

What Is NOT on This List

Three categories of business models that consistently underperform — not because they cannot generate revenue, but because they do not scale well enough to build venture-scale companies.

Consulting and Services

Every new customer requires more people. Revenue scales linearly with headcount, not exponentially with technology. Margins are constrained by labor costs. You cannot 10x revenue without approximately 10x-ing your team.

This does not mean consulting is a bad business. It means it is a bad startup business model if your goal is rapid, capital-efficient growth.

Hardware Only (No Subscription)

Requires heavy capital investment upfront — manufacturing, inventory, logistics, distribution. Margins are typically thin. There is no recurring revenue; every quarter starts from zero. Without a subscription or software layer, hardware companies face constant pressure to sell more units just to maintain revenue.

Built on Someone Else’s Platform

Building your entire business on top of another company’s platform — a Shopify app, a Salesforce integration, a social media API — creates existential platform risk. The platform can change its rules, raise its fees, or build your feature natively. And there is nothing you can do about it.

This does not mean you should avoid platforms entirely. It means your platform dependency should be a distribution channel, not your entire business.

The Recurring Revenue Advantage

One pattern dominates the most valuable companies: recurring revenue. Whether it comes from SaaS subscriptions, usage-based billing, or marketplace transaction fees, the companies that generate revenue predictably and repeatedly build more value than those that rely on one-time transactions.

The reasons are structural. Recurring revenue is predictable — you can forecast future revenue with reasonable accuracy, which makes planning and hiring possible. It compounds — each new customer adds to a growing base, and if your retention is strong, revenue grows even without new customer acquisition. And it creates higher customer lifetime value — a customer who pays $100/month for three years is worth $3,600, compared to a one-time purchase of $300.

Investors know this. SaaS companies have historically commanded a 20-40% valuation premium over comparable non-SaaS companies. One-time revenue streams are typically valued at 1-2x earnings, while recurring SaaS revenue commands multiples of 3-12x depending on growth rate and retention. The same dollar of revenue is worth dramatically more when it recurs.

5 Pricing Insights Every Founder Gets Wrong

1. You Should Charge

The most common pricing mistake is not charging at all. Free products attract users, but they attract the wrong kind of users — people who will use your product casually, provide lukewarm feedback, and disappear the moment you ask for payment.

Charging is a signal of value. If someone is willing to pay for your product, it means they value what you have built enough to trade their money for it. If nobody will pay, that is critical information — it means the value proposition is not strong enough, and you need to fix it before scaling.

As Aaron Epstein puts it in YC’s Startup School: pricing is a tool that helps you learn faster. You learn far more by charging than by giving your product away.

2. Price on Value, Not on Cost

Do not set prices based on what it costs you to deliver the product. Set prices based on the value you create for the customer.

If your software saves a company $100,000 per year in manual labor, charging $10,000 per year is reasonable — regardless of whether the product costs you $50 or $5,000 to operate. The customer is making a 10x return on their investment. They are happy to pay.

To find the value, talk to users and ask what problem they were hoping your product could solve. The answers almost always fall into four categories:

• Make more money — your product helps them increase revenue
• Save money — your product reduces their costs
• Save time — your product automates tasks they were doing manually
• Reduce risk — your product prevents costly mistakes or compliance failures

Each of these can be quantified. If you save someone 10 hours per week and their time is worth $50 per hour, that is $26,000 per year in value. Price accordingly.

3. You Are Almost Certainly Undercharging

The most common pricing mistake, after not charging at all, is charging too little. Founders are afraid to charge more because they think customers will leave. But if your product provides real value, customers will pay more than you expect. The discomfort you feel about raising prices is almost always greater than the resistance you will actually encounter.

A practical rule: keep raising prices until approximately 20% of prospects push back. If nobody is complaining about your price, you are leaving money on the table.

4. Pricing Is Not Permanent

Do not fear getting your pricing wrong the first time. You can always change it. Early-stage pricing is an experiment, not a commitment. Start somewhere reasonable, talk to customers, learn what value you provide, and adjust.

Grandfathering existing customers at their original price when you raise prices for new customers is standard practice and eliminates the risk of alienating your base. You can always go up; going down is harder because it signals a loss of value.

5. Keep It Simple

Complex pricing creates friction. Every additional tier, every additional feature toggle, every conditional discount adds cognitive load for the customer. The more decisions a customer has to make before purchasing, the more likely they are to defer the decision entirely.

The best pricing pages have two or three tiers that are easy to understand. Each tier serves a clearly different customer segment — individual, team, enterprise. The differences between tiers are obvious. There is no matrix of 43 features with checkmarks and X marks.

Simple pricing converts better because it requires less mental effort to evaluate.

The Segment Pricing Journey: A Case Study in Courage

Segment’s pricing evolution demonstrates every pricing insight simultaneously and remains one of the most instructive case studies in startup pricing history.

The company started by giving away its analytics product for free. Eventually the founders began charging $10 per month. Then something changed. A sales coach, brought in to help the team sell to larger companies, looked at Segment’s customer list and saw enterprise companies using the product. He insisted the founders quote $120,000 per year. The CEO resisted — it felt absurd for a product they had been giving away months earlier.

During their first enterprise sales meeting, the coach sat next to the CEO and threatened to quit if the CEO did not quote the full $120,000. When the customer asked for the price, the CEO reluctantly said $120,000. The customer countered at $12,000. They settled at $18,000 per year — still a 150x increase from the $10/month price point.

From there, Segment continued moving upmarket. Annual contracts grew to six figures. The company built a full enterprise sales motion. In 2020, Twilio acquired Segment for $3.2 billion.

The lesson is not simply to charge more. The lesson is that your understanding of your product’s value will evolve as you talk to customers, and your pricing must evolve with it. Segment’s founders could not have jumped from free to $18,000/year on day one — they needed the intermediate steps to understand what their product was actually worth to different customer segments. The founders who resist this evolution are the ones who leave enormous value on the table.

Sources & Further Reading

• Startup Business Models and Pricing — YC Startup School (YouTube)
• YC Guide to Business Models — Y Combinator Library
• Twilio Confirms Acquisition of Segment for $3.2B — TechCrunch
• 9 Business Models for Billion-Dollar Startups — Stone & Chalk
• Databricks Surpasses $4.8B Revenue Run-Rate at $134B Valuation — PR Newswire
• SaaS Gross Margin Benchmarks — CloudZero