Seat-based pricing is a measurement failure that hardened into a billing convention.
About 45% of SaaS is on hybrid pricing now, heading for around 60% by year-end. The teams moving fastest solved a measurement problem first, then changed the price. The shift to usage/outcome/agent metering gets framed as a pricing-strategy debate. The real blocker is instrumentation - pricing is the last thing that changes, after the measurement layer is in place.
Seat-based pricing is accurate when seats are a genuine proxy for the value the customer receives. A project management tool where more team members means more coordination handled is priced correctly on seats - the billing unit maps to the value unit reasonably well.
The gap appears when the product delivers value per outcome rather than per user: proposals generated, leads scored, documents processed, API calls resolved, agent tasks completed. Seats don't measure any of that. They measure access, not delivery.
AI-native SaaS app spend more than doubled year-over-year. That's the fastest-growing pricing segment by a wide margin, and the reason is structural: AI-native products ship with instrumentable outcomes at build time. You can count the tokens consumed, the tasks completed, the hours of labor replaced. The measurement layer that justifies usage-based billing is part of the product from day one, because the product has to track usage to know whether it's working at all.
For SaaS founders still on seat pricing, the diagnostic I'd run is one question: can you instrument the unit of value your customer actually receives at the per-customer level? If you can't measure it, you can't price on it. Most seat-based SaaS has never built that instrumentation because seats were good enough and the engineering time went elsewhere. That's a reasonable historical choice. It becomes a competitive problem when AI-native competitors are pricing on the outcome your customers actually care about.
The attribution parallel for SaaS UA is direct. Billing on seats when your product delivers per-outcome value is the same category of signal error as bidding Google on demo-count when what you want to optimize is pipeline-value-weighted closed-won revenue. The billing unit - like the conversion event - determines what the system optimizes toward. Both are wrong in the same structural way.
Getting the measurement layer right is the prerequisite - for the pricing model and for the bidding signal.
Ruslan co-founded Tegra in 2017. Runs the Google Ads practice - feed, PMax, search, attribution. Writes weekly about the parts of paid search operators are afraid to touch.