How do you price a product?

Three approaches, only one of which is usually right

Cost-plus — compute your costs, add a margin. Defensible for commodities and almost meaningless for software, where marginal cost approaches zero and the price would be arbitrary anyway.

Competitive — anchor to what similar products charge. Fast, and it hands the most powerful lever in your business to a company whose costs, customers and strategy you do not know.

Value-based — price against the value the customer receives. Harder, requires actually understanding your buyer, and is the only one of the three that can capture what you are worth.

Structure before level

Two decisions, in order.

The value metric. What you charge by. The test is whether it rises as the customer gets more value. Seats work when value scales with team size. Transactions work when value scales with volume. A flat monthly fee works when neither varies much, and it caps you at the value of your smallest customer.

Stripe's percentage-of-transaction model is the clean case: a customer processing more money is getting more value, and pays proportionally, with no renegotiation and no tier cliff.

The level. Once the metric is right, the number is a narrower question — and one you can test.

Packaging is where segments get separated

Tiers exist to let different segments self-select. The mistake is building tiers around arbitrary feature groupings rather than around what distinct segments actually need.

Canva's freemium split works because the free tier delivers genuine value to individuals, and the paid tier is organised around things teams and businesses need — brand consistency, collaboration, asset management. The line falls where the segments naturally divide, so upgrading feels like outgrowing rather than being blocked.

The failing version withholds something the free tier obviously ought to have. Users experience that as hostility, and it poisons the upgrade rather than motivating it.

Why most products are underpriced

Founders and PMs anchor on their own willingness to pay, which is systematically lower than that of a business buyer solving an expensive problem. A tool saving a team twenty hours a month is not competing with other tools' prices; it is competing with the cost of twenty hours.

Zerodha ran the opposite play deliberately — a flat, radically low fee that was a strategic weapon rather than an underestimation, aimed at a segment incumbents could not profitably serve. That works when low price is the strategy. It is very different from a low price arrived at through nervousness.

Testing before committing

Ask existing customers what they would do at a higher price and you will get a hypothetical. Better signals:

• Raise the price for new customers only and watch conversion, not opinion.
• Offer an annual prepay discount and see who takes it — a proxy for confidence in the product.
• Run price differences by segment or geography where legitimate.
• Watch which tier people actually pick when the middle option changes.

And revisit it. Pricing set at launch and never touched is pricing set by a company that no longer exists, for a product that has since become more valuable.