Startup Revenue Models Explained: 8 Models and How to Choose

Startup revenue models describe how a company turns what it builds into cash: subscriptions, usage-based fees, one-time transactions, marketplace take rates, licensing, advertising, freemium conversion, or a hybrid. In our view, a good model matches how your customer receives value, how often they receive it, and how they prefer to pay.

Many durable startups settle on one primary model that produces most of their revenue, with one or two secondary models layered on later.

Below, each model in plain terms, with the metrics investors will ask about, where it fits and what to watch for, followed by a five-question framework you can use as one input when choosing.

What a revenue model is (and how it differs from pricing)

A revenue model answers "what do we charge for, and how does the money flow?" Pricing answers "how much?"

Change your price and revenue moves by a percentage. Change your model and you can change your customer base, sales motion, cash flow timing and how investors value the company.

A quick test: if you can't describe your revenue model in one sentence ("we charge a monthly fee per seat," "we take a percentage of every booking"), you may not have chosen one yet.

The eight startup revenue models

1. Subscription

Customers pay a recurring fee, monthly or annually, for ongoing access. It is the common default for software because it is predictable and compounds.

Metrics that matter: monthly and annual recurring revenue (MRR, ARR), gross and net revenue retention (GRR, NRR), churn, customer acquisition cost (CAC) payback and lifetime value to CAC.

Benchmarks vary by segment and contract size. ChartMogul's 2023 SaaS Retention Report puts best-in-class NRR in the 110 to 120 percent range, but it also found that even top-quartile companies with $1M to $3M of ARR had NRR around 94 percent. Retention is much stronger at higher price points: only 2.7 percent of businesses with average revenue per account under $10 a month had NRR above 100 percent, versus 41.1 percent of those above $500 a month. Compare yourself with companies of similar size and price point. Our guide to startup business models and unit economics shows how to calculate CAC, LTV and payback.

Fits: B2B tools, consumer apps with ongoing utility, content and media, physical replenishment subscriptions.

Watch for: charging a subscription for something the customer uses twice a year. How often you deliver value should justify how often you bill.

2. Usage-based (consumption)

Customers pay for what they consume: API calls, compute, messages, documents processed. Revenue tracks value closely and grows as the customer grows.

Metrics that matter: usage growth per account, gross margin per unit (critical when your own costs are usage-based, as with AI inference; see AI gross margins and inference costs), net revenue retention and revenue concentration.

Fits: infrastructure, developer tools, AI products, communications and payments.

Watch for: unpredictable bills that make buyers nervous. Many usage-based companies add committed tiers or a base fee to smooth revenue, which is one reason hybrid pricing is spreading. Kyle Poyar's 2025 State of B2B Monetization survey of more than 240 software and AI companies found hybrid pricing rose from 27 percent to 41 percent of respondents over the prior year.

3. Transactional (one-time sales)

Customers pay once for a product or service: hardware, a course, a consulting engagement, a physical good. Easy to understand, but revenue has to be re-earned with each sale.

Metrics that matter: gross margin, average order value, repeat purchase rate, CAC per order and contribution margin after fulfillment and returns.

Fits: e-commerce and DTC brands, hardware, services and events.

Watch for: growth that requires acquisition spend to rise in lockstep. Strong transactional businesses often find a repeat loop (replenishment, upgrades, community) that starts to behave like a subscription.

4. Marketplace take rate

The platform connects buyers and sellers and keeps a percentage of each transaction. It owns matching, trust and payments, not the inventory.

Metrics that matter: gross merchandise value (GMV), take rate, net revenue (GMV times take rate), liquidity (how often a listing sells or a request is filled), concentration and repeat rates on both sides. As a16z's guide to marketplace metrics puts it, GMV does not equal revenue.

Take rates vary widely by category and by how much work the platform does. For a current reference point, Etsy's fee policy charges sellers a 6.5 percent transaction fee plus a $0.20 listing fee, before payment processing and optional ad fees.

In his 2013 essay "A Rake Too Far," Bill Gurley put eBay's rake at roughly 10 percent and Amazon's marketplace at roughly 12 percent (6 to 15 percent by category), against 30 percent for Apple's app store, and argued that a modest rake on high volume is the more durable position. As a rule of thumb, platforms that do more of the work can charge more, while very large transactions tend to support only small percentages.

Fits: markets with fragmented supply and demand and a trust or discovery problem.

Watch for: disintermediation (buyers and sellers meeting once, then going direct) and the cold-start problem. Marketplaces often need to subsidize one side early, which can be expensive.

5. Licensing

Customers pay for the right to use your technology, data or brand, often as a fixed annual fee or a royalty. Common when the customer wants to run the product themselves or embed it in their own.

Metrics that matter: annual contract value, renewal rate, sales cycle length and customer concentration.

Fits: deep tech, datasets, models, IP-heavy products and brands with licensing potential.

Watch for: long sales cycles and dependence on a few large deals. Investors tend to discount lumpy revenue unless renewals are proven.

6. Advertising

Users get the product free and advertisers pay for their attention.

Metrics that matter: daily and monthly active users, time spent, average revenue per user (ARPU), fill rate and effective CPM.

Fits: media, communities and consumer apps with very large audiences.

Watch for: needing a very large, engaged audience before ads produce meaningful revenue. As a primary model, advertising can be hard to fund at seed unless the audience is already big and growing organically.

7. Freemium

A free tier drives adoption; a paid tier converts a fraction of users. Strictly, freemium is a distribution strategy paired with a subscription or usage model, but founders treat it as a model, so it belongs here.

Metrics that matter: free-to-paid conversion, time to conversion, cost to serve free users and the specific limit or feature that triggers an upgrade.

Conversion varies enormously. In Kyle Poyar's 2026 Free-to-Paid Conversion Report, covering 200 software products, one in four freemium products converted below 2.5 percent, conversion above 15 percent was rare, and the median across all products surveyed (free trials and freemium together) was 8 percent.

Fits: products with viral or collaborative loops where free users bring in paying ones.

Watch for: a free tier so generous nobody needs to pay, or so stingy nobody sticks around. The best upgrade trigger, in our view, is a moment of real value, not frustration.

8. Hybrid

Many scaled companies mix models. In software, the usual pattern is the one Stripe's guide to SaaS pricing describes: a flat recurring fee combined with variable usage charges. Other mixes include a marketplace take rate plus premium seller subscriptions, or a DTC brand with a replenishment plan. Hybrids capture more value but add complexity to pricing, billing and reporting.

Metrics that matter: revenue mix by model and its trend, blended gross margin, and whether the secondary model improves or dilutes retention.

Fits: companies past initial product-market fit with a clear primary model and an adjacent way to capture value.

Watch for: adopting a hybrid before the primary model works. Two half-working models are usually worse than one working model.

"More revenue streams means less risk"

It sounds prudent. Diversify early, don't depend on one way of making money, give investors more than one story.

But at the early stage, each model needs its own pricing, its own sales motion and its own metrics. Run three at once and you learn slowly about all of them. We'd pick one, make it work, and add a second only when the first is clearly earning. And as we've argued before, not all revenue is good revenue; some of it quietly costs more than it brings in.

How to choose among startup revenue models: a five-question framework

Work through these in order with your best customers in mind. Treat the answers as a guide, not a formula.

How does the customer experience value? Continuously (subscription), in proportion to use (usage), in discrete moments (transactional), or through a match with someone else (marketplace)?
How does the customer prefer to pay? Ask them. Finance teams often like predictable bills; developers like paying for what they use; consumers like low-commitment entry points.
What does your cost structure look like? If costs scale with usage, it often makes sense for revenue to scale too. If costs are mostly fixed, subscriptions can maximize margin.
What sales motion can you run? Usage and freemium tend to suit product-led growth. Licensing and large subscriptions usually need founder-led, then rep-led, sales.
What will investors need to see? Recurring revenue with strong retention and gross margin generally earns higher valuation multiples; transactional and advertising businesses are judged more on growth and margin, and the bar is often higher. See how exit multiples connect revenue quality to value.

Then write down the answers, pick one primary model, and give it a couple of quarters before evaluating a change.

Worked example: one product, three models

An illustrative example: suppose you sell an AI tool that drafts insurance claim summaries. The prices below are hypothetical.

Subscription: $500 per seat per month. Predictable, but a 5-person team pays $2,500 a month whether it processes 2,000 claims or 200.
Usage: $2 per claim summary. A team processing 2,000 claims a month pays $4,000; revenue tracks value, but monthly bills swing.
Hybrid: a $1,500 monthly platform fee that includes 500 summaries, then $1.50 per extra summary. The same 2,000-claim team pays $3,750, finance gets a predictable floor, and you keep upside as volume grows.

Check gross margin in each case. If each summary costs you $0.60 in inference, your margin per summary is healthy under usage pricing but can shrink under a flat seat price for heavy users.

Common mistakes with startup revenue models

Choosing the fashionable model rather than the one customers will pay for.
Launching with three models at once and learning nothing about any of them.
Ignoring gross margin. Bessemer's State of AI 2025 found its fastest-growing AI "Supernovas" averaged about 25 percent gross margin (often negative), against about 60 percent for steadier "Shooting Stars," both below typical SaaS. Thin margins tend to be valued differently until they improve.
Confusing bookings, revenue and cash. A $120,000 annual contract billed monthly is $10,000 of revenue this month.
Not revisiting the model. As the product and customer base evolve, the way you charge may need to change too; we see pricing as an ongoing process, not a launch decision. Our guide on how to price your product covers the levers within a model, and the startup KPI framework shows which metrics to track at each stage.

Where 1752vc fits

Choosing and validating a revenue model is a foundational step in building an investor-ready business, and one many founders skip until an investor asks "how do you make money?" 1752vc's Ignite, a 12-week startup academy for first-time founders in the early stages, is one structured place to work through questions like this while the MVP is still taking shape. Consumer brands weighing subscription and transactional models can look at Ignite DTC, which focuses on AI-driven marketing and omnichannel growth. Once the model works and you are ready to raise, the startup fundraising guide covers what comes next.

The bottom line

Charge the way your customer already experiences value, prove one model before you stack another, and watch the margin as closely as the top line.

The price is a number you can change next week.

The model is a bet on how your customer thinks.

Key takeaways

Startup revenue models define what you charge for and how cash flows; pricing decides how much, and the model is, in our view, the bigger decision.
Subscription and usage models tend to earn strong investor confidence when retention and gross margins hold; benchmark retention against companies of similar size and price point.
Hybrid subscription-plus-usage pricing is spreading quickly, rising from 27 to 41 percent of surveyed software and AI companies in the year to mid-2025, per Kyle Poyar.
Marketplaces live on GMV, take rate and liquidity, and take rates vary widely by category (Etsy's transaction fee, for example, is 6.5 percent).
Freemium tends to work best when the upgrade trigger is a moment of real value; Poyar found a quarter of freemium products convert below 2.5 percent and conversion above 15 percent is rare.

Frequently asked questions

What are the main startup revenue models?

The most common startup revenue models are subscription, usage-based, transactional (one-time sales), marketplace take rate, licensing, advertising and freemium, with many companies running a hybrid of two. Each has a distinct sales motion, cash flow pattern and set of metrics that investors evaluate.

Which revenue model is best for a startup?

We do not think there is a single best model. A good fit matches how your customer receives value, how they prefer to pay, and how your costs scale. Subscription and usage-based models are common in software because they produce recurring revenue, while marketplaces and transactional models fit businesses built on matching or physical goods.

What is a good take rate for a marketplace startup?

It depends on the category and how much value the platform adds. Goods marketplaces commonly sit between mid single digits and the low teens: Etsy's transaction fee is 6.5 percent, and Bill Gurley's 2013 analysis put eBay near 10 percent and Amazon's marketplace near 12 percent. Platforms that do more of the work can often charge more, but an overly greedy rake can invite competition.

What is a good freemium conversion rate?

It varies widely. Kyle Poyar's 2026 Free-to-Paid Conversion Report, covering 200 software products, found one in four freemium products convert below 2.5 percent and conversion above 15 percent is rare. In our view, the trend in your own conversion and the specific trigger that drives upgrades matter more than a single benchmark.

Can a startup have more than one revenue model?

Yes, and many scaled companies do, most often a subscription plus usage fees. Many early-stage founders find it helps to establish one primary model first, prove it with real retention and margin data, and only then add a secondary model that captures adjacent value without confusing customers or complicating billing.

Sources

Disclaimer: This guide is for general education only and is not legal, tax or investment advice. Laws, market data and program terms change, so it may not reflect the latest developments or fit your situation. Treat it as a starting point, not a source of truth, and talk to a qualified lawyer, accountant or financial adviser before you make decisions.