Usage-based vs seat-based pricing: 5 models and when each works

Seat-based pricing charges for each person with access. Usage-based pricing charges for what the product does: API calls, compute, credits, transactions, or outcomes. Seats work when value grows with headcount. Usage works when value and cost grow with activity. Many SaaS and AI companies end up with a hybrid of the two.

The harder question is which usage model to use. There are five common ones, and each fits a different kind of product.

What are the most common pay-as-you-go pricing models used by SaaS companies?

The five models below cover most usage-based pricing in SaaS and AI. Examples were checked against each company's public pricing page in October 2026.

A few patterns sit behind that table.

Pay-per-use ties the bill directly to consumption, and it shows up more in infrastructure than in application software. Customers never pay for idle capacity, and your revenue tracks your costs. The trade-off is forecasting. Revenue moves with your customers' activity, in both directions.

Credits belong to the user, not to the infrastructure. Credits work when a person starts each unit of work: generating an image, enriching a lead, running a research task. If the balance runs out, the user waits or tops up, and nothing breaks. Infrastructure is different. Nobody wants a production database or an inference endpoint to stop because a credit balance hit zero. For background, business-critical workloads, charge automatically and protect the customer in other ways.

Percentage pricing needs visibility into money. Payments, marketplaces, and creator platforms can do it because they process the transaction. Most software can't see its customers' revenue, so the model rarely fits.

Outcome-based pricing is newer in software than in services. Recruiters and commission-only agencies have priced on results for decades. AI makes it possible for software when the product does the work end to end and the result is easy to verify.

Usage-based pricing vs subscription: what actually changes?

A subscription is a fixed fee for a period. Seat-based pricing is the most common subscription unit in B2B SaaS: a price per user per month.

Usage-based pricing changes three things.

1. What scales the bill. Headcount for seats, activity for usage.
2. Who carries the cost risk. With unlimited usage on a fixed seat price, the vendor absorbs heavy users. With usage pricing, cost follows consumption.
3. What you need to bill. Seats need a count at renewal. Usage needs metering, aggregation, and pricing logic running every billing period.

Usage-based pricing isn't new or fringe. In OpenView's 2021 State of Usage-Based Pricing survey of about 600 SaaS companies, 45% reported using it, up from 34% the year before.

Neither model is better in general. Seats fit collaboration tools, where more people using the product is the value. Usage fits products where software, not people, generates the activity, and where each unit of activity carries a real cost.

Is seat-based or usage-based billing better for net revenue retention?

Neither guarantees higher NRR. Each ties expansion and contraction to a different driver.

With seats, NRR follows your customers' headcount. You expand when teams hire and contract when they cut. You don't capture more revenue when one user gets ten times more value. And if your product automates work, it can reduce the number of seats a customer needs.

With usage, NRR follows your customers' activity. You expand automatically as customers use more, without a renegotiation. OpenView's 2021 research reported that usage-based companies had higher net dollar retention than their peers. The same mechanism works in reverse. Snowflake, a consumption-priced data platform, reported a net revenue retention rate of 178% as of January 31, 2022, and 131% as of January 31, 2024. Both are strong numbers. The gap is a reminder that usage-based NRR moves with consumption, in both directions.

For a B2B SaaS company, the practical answer is usually:

• Keep a committed or recurring component so contraction has a floor.
• Add a usage component that expands with the value customers get.
• Price the usage on a metric the customer can understand and influence.

That last point is often where a pricing model succeeds or fails. The playbook includes a four-question test for choosing a billable metric, with worked examples. Get the full test in the usage-based pricing playbook.

Is consumption-based or seat pricing better for AI chatbots?

For most commercial AI chatbots, pure seat pricing is a poor fit, because the cost and the value sit in the conversations, not in the logins.

Every AI conversation consumes inference. A single seat can trigger thousands of conversations, and an AI agent that answers customers directly may not need a seat at all. A flat per-seat price leaves the vendor exposed to its heaviest users.

Three options work better:

• Consumption: charge per conversation, message, or token. Simple and cost-aligned, but harder for buyers to budget.
• Outcome: charge per resolved conversation. Intercom prices its Fin AI agent from $0.99 per outcome, counted when the customer confirms the issue is resolved, doesn't ask for more help, or Fin completes a workflow.
• Hybrid: keep seats for the humans and add a usage or outcome charge for the AI. Intercom does this too: per-seat plans for the support team, plus per-outcome pricing for Fin.

If the chatbot is a self-serve tool that users start themselves, credits also work. If it's part of a customer's production support flow, it should keep running and charge automatically.

What is hybrid billing?

Hybrid billing combines a recurring fixed fee with usage-based charges on the same subscription. A typical hybrid plan has three parts:

• A base fee or seats, billed in advance, which gives you predictable revenue.
• Included usage, which gives the customer a clear starting point.
• Overages or top-ups, billed on actual consumption beyond the allowance.

Supabase is a clear example. Its Pro plan has a monthly fee that includes a quota of monthly active users, database size, storage, and egress. Usage beyond the quota is billed per unit, and a spend cap is on by default.

A common operational risk in hybrid billing is overages that pile up between billing cycles. An AI customer can run up a large inference bill in days, then dispute it or churn before paying. Progressive billing handles this. You set spending thresholds, track cumulative usage, and issue an invoice automatically when a threshold is reached, instead of waiting for the end of the period. This protects your cash flow and stops customers from building up debt they didn't expect.

What to look for in a usage-based pricing platform for SaaS

The right platform can bill every model above without custom code, so you can change pricing without rebuilding billing. Check for:

• Plans that combine fixed and usage charges. In Lago, a plan can carry a base subscription fee, fixed charges such as seat licenses, and usage-based charges on billable metrics, all on one subscription.
• Several charge models. Lago supports standard, graduated, package, volume, percentage, graduated percentage, and dynamic pricing. See the charge models documentation. Per-transaction minimums and maximums on percentage charges are a premium feature.
• Prepaid credits and wallets. Lago wallets hold prepaid and free credits, apply them to invoices automatically, and support recurring top-ups triggered on an interval or when the balance drops below a threshold. Real-time wallet balance is a premium feature.
• Overage protection. Progressive billing, available in Lago as a premium add-on, invoices customers when lifetime usage crosses thresholds you define.
• Deployment choice. Lago is open source and can run self-hosted or on Lago Cloud.

Pricing also has to be visible to the customer. Usage pricing without dashboards, alerts, and clear invoices produces surprise bills and churn. The playbook includes a checklist of the customer-facing features to build alongside the model.

To see how this works for usage-based plans, visit Lago for usage-based billing.

Choosing your model

Start from two questions. Does value grow with people or with activity? Is the workload user-initiated or business-critical? The answers point you to seats, credits, automatic usage charges, or a hybrid.

The playbook covers the rest: choosing a billable metric, setting included usage and overages, prepaid versus postpaid, and the features customers need to trust a usage-based bill.

Download the usage-based pricing playbook

Want to model your own pricing in Lago? Book a demo.