Usage-Based Billing for SaaS: When to Use It and How to Ship It

Usage-based pricing went from a Snowflake curiosity to the default expectation for AI products in about eighteen months. Founders now ask for it by name. But ‘charge per use’ hides a surprising amount of engineering, and for plenty of products it's the wrong model entirely. Here's how to decide, and what it costs to build.

The three models, and who each one is for

• Flat-rate: one price, all features. Best for early MVPs, simplest to build and to explain.
• Per-seat: price × number of users. Best for collaboration tools where value scales with the team.
• Usage-based: price × consumption (API calls, tokens, GB, messages). Best when your own costs scale with usage, classic for AI.

The honest rule: charge the way your costs behave

If a heavy customer costs you almost nothing more than a light one (a project-management tool), flat or per-seat pricing protects your margin and your sanity. If a heavy customer burns real money, every request hits an LLM or a third-party API, usage-based pricing isn't a growth hack, it's survival. Mispricing here is how AI startups post negative gross margins.

“Pick the pricing model that makes your most enthusiastic customer your most profitable one, not your most expensive.”

What metered billing actually requires you to build

This is the part founders underestimate. ‘Just charge per use’ expands into a real subsystem:

1. A metering pipeline. Every billable event recorded reliably, idempotently, without slowing the request that triggered it.
2. Aggregation. Rolling those events into per-customer, per-period totals you can trust enough to invoice.
3. Stripe Billing Meters. Reporting usage to Stripe and reconciling when your numbers and theirs disagree.
4. An in-app usage view. Customers will not tolerate a surprise invoice; show live consumption and a projected bill.
5. Limits and alerts. Soft caps, hard caps, and ‘you're at 80%’ emails so nobody gets a shock.

Don't make customers do mental math

Pure usage pricing terrifies buyers because they can't predict the bill. The fix most winning products use in 2026 is hybrid pricing: a flat platform fee that includes a generous usage bucket, then metered overage above it. Predictable for the customer, margin-safe for you. It's the best of both, and it's what we recommend to most SaaS founders adding usage today.

When NOT to do usage-based billing

If you're pre-revenue and still proving people want the thing, skip it. A flat $29/month plan you can ship in a day will teach you more about willingness-to-pay than a perfect metering system will. Add usage-based pricing when you have paying customers and a clear, cost that scales with their consumption, not before.