Pricing for early startups: why you are probably charging too little
Ask a room of early founders how they set their price and the honest answers are: guessed, copied a competitor, or picked what felt safe. All three converge on the same error. First prices are almost always too low, because founders price their own fear of rejection rather than the customer's cost of the problem.
The stakes are not abstract. A much-cited McKinsey analysis of large-company economics found that a 1 percent improvement in price, volume and costs held constant, lifts operating profit by roughly 11 percent — more than the equivalent improvement in volume or cost. Small companies are not large ones, but the direction of the leverage is the same, and early startups leave it untouched.
Why founders underprice
Underpricing feels risk-free: more people say yes, and each yes is a small hit of validation. But a low price does three kinds of damage that are hard to see from inside. It selects for the least committed customers, who churn fastest and complain loudest. It starves the channels you will need later, because a $9 product cannot support sales calls or paid acquisition. And it anchors the market's sense of what the product is worth, which is painful to raise later.
The counterintuitive finding from founders who publish their experiments: doubling a too-low price typically cuts signups by far less than half. A product that loses 20 percent of signups at twice the price just grew revenue 60 percent with fewer customers to support.
Pick a value metric
Before the number comes the unit: what exactly the customer pays for. The value metric is the thing that grows as the customer gets more value — seats for a team tool, contacts for an email platform, transactions for a payments product. A good value metric does three jobs: it is easy to understand before purchase, it scales with the customer's success, and it expands revenue without a renegotiation when the customer grows.
- Charge for the unit the customer already counts: users, orders, projects, gigabytes.
- Avoid pricing on your costs; the customer does not care what your servers cost.
- Keep three tiers maximum; a fourth tier is usually a fear, not a plan.
- Grandfather existing customers when you raise prices; new prices are for new decisions.
Testing price without burning trust
Price research has one reliable instrument and several dishonest ones. Surveys that ask "what would you pay?" return fiction. The Van Westendorp method asks four questions — at what price is this too cheap to trust, a bargain, getting expensive, and too expensive to consider — and the intersection of the curves gives a defensible range rather than a vibe. For a live product, the cleaner test is behavioral: show the new price to new visitors only, and read the conversion and refund data after a few hundred checkouts.
| Model | Best for | Main risk | 2026 example pattern |
|---|---|---|---|
| Flat monthly fee | Simple tools, one user type | Leaves money on big accounts | $29–99/mo indie SaaS |
| Per seat | Team collaboration products | Punishes adoption inside a company | $10–30/user/mo work tools |
| Usage-based | Infrastructure, APIs, AI tools | Unpredictable bills scare buyers | Per 1,000 calls or credits |
| Freemium | Products with viral spread | Free users consume support forever | Free tier capped at real usage |
| Annual upfront | Anything with proven retention | Hides churn for a year | Two months free, cash now |
When freemium is a strategy and when it is a subsidy
Freemium works when free users create value for paying ones — through network effects, word of mouth or content — and when the marginal cost of a free user is near zero. It fails as a default answer to "nobody is buying." If the paid plan does not convert, a free plan converts the problem into a larger, quieter one. Fix the value first, then decide whether giving some of it away is a growth engine or a charity.