Metronome argues token billing is poor customer-facing pricing, advocates for unified credits and output-based models

Token billing is useful infrastructure for tracking usage and managing markups, but it is a poor customer-facing pricing model. Charging based on token consumption lets costs determine price, positioning the product as a commodity markup rather than a value provider. When invoices reveal model mixes and margins, customers can challenge the markup or route around the provider as models become cheaper and interchangeable.

Unified Credits as a Solution

Unified credits offer a presentation layer that translates underlying token usage into tangible operations for the customer. Instead of seeing token counts and model costs, customers see credits consumed by specific tasks, such as enriching a dataset or generating an image. This approach keeps the complexity of model routing and margin tuning behind the scenes while allowing companies to maintain control over their internal economics. Metronome views this as a stepping stone to output-based pricing.

Output-Based Pricing vs. Outcome-Based Pricing

While output-based pricing is the ideal alignment of AI consumption with value, outcome-based pricing is largely a myth for most companies. Outcome-based pricing requires monopoly-like power or very high ticket prices to justify the investment in measuring and attributing results like revenue growth or churn reduction. In contrast, outputs are objective, verifiable metrics like resolved support tickets or generated images. Companies should define their output metric and market it as an outcome to align pricing with value.

The Future of AI Pricing

Pricing models shift rapidly once a new standard gains traction in a mature market. Hybrid pricing (seats plus usage) saw adoption by roughly one in six Stripe users crossing key revenue milestones as of August 2026, after years of low usage. Similarly, Fin’s adoption of pricing based on resolved support conversations quickly led all five leading companies in the AI customer-support category to commit to some version of outcome-based pricing. As AI agents increasingly evaluate and purchase services, pricing will likely shift toward measurable outputs, unbounded by the complexity limits of human buyers.