The short version: By usage. Unlike SaaS, additional users and action aren’t marginal cost—they are significant. In the videos below, Everett Berry, Head of GTM Engineering at Clay, explains how they use a system of credits and tiers.
Unlike traditional SaaS, compute costs rise with usage and can eat into your margins. As just one example, a single research query allegedly costs OpenAI $55. At the same time, AI does real work. It completes full tasks and can own workflows. This means AI startups have the opportunity to bill for the value of the work done.
How you charge depends on how clearly you can measure, report on, and convey that value. This is easier from some AI startups than others. Clay has found success with a system of credits where, instead of charging for overages, customers must upgrade to a higher tier which also includes additional functionality and integrations.
In the following video, Everett explains how they arrive at this model, and why its genius is the fact that it aligns Clay’s cost with the value it delivers.
Watch: Everett on Clay’s early model and customer value
Unlike pure usage-based SaaS, Clay intentionally throttles usage to prompt a sales conversation. This guardrail prevents customers who don’t understand the platform from buying far more than needed, never adopting, and churning.
As Everett explains, a customer buying millions and millions of credits might not actually know what they’re doing. It’s a great opportunity for the Clay team to educate them.
Watch: Everett on limiting app usage
Clay has learned that the simplest pricing is the best. They used to charge a platform fee in addition to their credits but “procurement teams would rip it apart” and clients would spin out trying to figure out what “value” they were getting from the software. Whereas if Clay simply bundled the platform fee within the credits, customers were happy to pay.
Watch: Everett on why Clay is all-in on credits