How it works
You pick a value metric that tracks what customers get and what you pay to serve it: API calls, tokens, minutes, gigabytes, or completed tasks. Each unit has a price, sometimes with volume discounts. Revenue then grows with adoption instead of being fixed at a seat count.
The billing model decides when money moves. Metered billing records usage and invoices it at the end of the period. Prepaid credit billing sells the usage first and deducts it as it happens. Many products blend the two with a subscription that includes a usage allowance.
Example
An image API charges $0.04 per image. A customer who generates 2,500 images in May pays $100. Under metered billing they receive a $100 invoice on June 1. Under prepaid credits they would have bought, say, a 3,000-credit pack for $120 and still hold 500 credits.
Usage-based pricing in Cred
cred.diy implements the prepaid form: usage is a credit deduction against a balance the customer already paid for, so there is no end-of-month invoice to collect. Subscriptions with a monthly credit allowance cover the hybrid case.
Managing Credits docsQuestions
- What is a good value metric?
- One the customer understands, that grows as they get more value, and that roughly follows your cost to serve. Tokens, generations, and API calls work well for AI products because each maps to a model cost.
- Is usage-based pricing the same as pay-as-you-go?
- Pay-as-you-go is one form of it. Usage-based pricing also includes committed-use contracts, tiered volume pricing, and subscriptions with included usage plus overage.
- How do I move from flat pricing to usage-based pricing?
- Usually gradually: keep existing plans, add an included usage amount to each, and charge for usage above it. That lets you learn real usage patterns before setting final per-unit prices, and gives current customers time to adjust.
