The per-seat pricing page is a relic

Look at the pricing page of almost any software company and you will see the same architecture: a few named tiers, each priced per user, per month. That structure is so universal it reads as a law of nature. It is not. It is an artifact of one specific assumption — that software is a thing humans log into — and that assumption is now breaking.

Why per-seat made sense

Per-seat pricing won because it was a clean proxy for value. More employees using the tool meant more value extracted from it, so charging per employee tracked value reasonably well. It was also legible: a buyer could forecast the bill, and a vendor could forecast revenue, both just by counting headcount.

The whole SaaS go-to-market machine got built on top of that proxy. Sales comp, expansion targets, the definition of "net revenue retention" — all of it assumes a seat is the unit you sell.

What agents break

An agent does not have a seat. It does not log in in the morning. One deployed agent might handle the work of fifty people, or five hundred, or — on a quiet week — almost none. The relationship between "number of agents" and "value delivered" is not loose. It is nonexistent. Counting agent instances tells you nothing.

So the proxy is gone, and a vendor who keeps the per-seat page is now in one of two bad spots. Either they price the agent like a very expensive employee — and watch a buyer rationally deploy one agent where they'd have bought ten seats, collapsing the account. Or they invent a fake "seat" for the agent and hope nobody notices that the unit no longer corresponds to anything.

This isn't a far-off problem. It's visible right now in every expansion forecast built on seat growth, in a market where the customer's next unit of work is increasingly done by an agent that was never going to be a seat.

What replaces it

Pricing has to re-attach to something real. The candidates, roughly in order of how directly they track value:

• Outcome-based. Charge per resolved ticket, recovered charge, completed task, closed loop. The unit is the thing the customer actually wanted. Hardest to meter cleanly — you have to define and verify the outcome — but it aligns vendor and customer incentives almost perfectly.
• Consumption-based. Charge for work performed — actions taken, runs executed, tokens if you must. Easier to meter, but it prices effort, not value, and an agent that does more work to reach the same result costs the customer more. That's a subtle misalignment, and customers eventually feel it.
• Hybrid. A platform fee for being deployed, plus an outcome or consumption component on top. This is where most serious agentic pricing is landing, because the platform fee gives the vendor a predictable floor and the variable component keeps the bill honest as usage swings.

We worked through the economics in Pricing Agentic Outcomes. The headline: per-seat doesn't survive the transition, and the replacement that holds up is hybrid with a genuine outcome component — because outcome is the only unit that still means something when the worker isn't a person.

The hard part isn't the model. It's the migration.

Most teams reading this already suspect per-seat is on the way out. What stops them isn't disagreement — it's the switching cost. The CRM, the comp plan, the board metrics, the renewal contracts: all of it is denominated in seats. Re-denominating a whole company in outcomes is a genuine project, and "our pricing page is technically outdated" rarely beats the things screaming louder this quarter.

So here is the realistic move. You do not need to rip out per-seat next week. You need to start metering the outcome now — instrument what your product actually accomplishes for a customer, even while you still bill by seat. The day the seat model visibly stops working for a real account, the company that has been quietly measuring outcomes for a year can change its pricing page in an afternoon. The company that hasn't is starting a year-long project under pressure.

The pricing page is the last room in the SaaS house to get the renovation. It is also the one the customer reads first.