Skip to content

7 · Pricing an AI product or service

Pricing is one of the highest-leverage decisions you'll make — and AI products have a twist most don't: a real, per-use cost (Lesson 6). Price too low and a busy month loses you money; price on cost alone and you leave value on the table. Here's how founders think about it.

Two honest pricing approaches (and why you usually blend them):

  • Cost-plus. Start from what it costs you to serve a customer — including your AI cost-to-serve — and add a margin. This is your floor: a price below your cost-to-serve means every sale loses money. With usage-based AI costs, the floor is non-obvious, which is exactly why Lesson 6 came first.
  • Value-based. Price on what the outcome is worth to the customer — the time, money, or pain you save them — not on your cost. A report that saves a customer 10 hours is worth far more than the few cents of tokens it took to draft. This is usually where the real price lives.

The honest rule: your cost-to-serve sets the floor; the value you deliver sets the ceiling; your price lives in between, above the floor. The SBA's business-plan and startup-cost guidance push the same discipline — know your costs, tie price to financial outcomes, and make sure the numbers actually work (SBA, n.d.).

Common AI-product pricing models:

ModelGood when…Watch out for…
Subscription (flat monthly)Steady, predictable useA heavy user whose AI cost exceeds their fee
Usage-based (pay per use/credit)Use varies a lot per customerCustomers fearing a surprise bill
Tiered (good/better/best)Different customer sizesToo many confusing tiers
One-time / per-project (services)A delivered outcomeUnderpricing your own time

Founder cautions specific to AI:

  • Watch the heavy user. Flat pricing + usage-based AI cost = one power user can wipe out your margin. Cap usage, or move heavy users to a higher tier.
  • Don't race to the bottom. "Cheaper than the big AI company" is not a strategy; better for this specific customer is.
  • Be honest about value. Overstating what your AI does to justify a price isn't just bad pricing — overstated performance claims can be illegal (Lesson 8). Charge for real value, described truthfully.

Check yourself. What sets the floor on your price, what sets the ceiling, and why is flat-rate pricing risky when your AI cost is usage-based?

Sources