Seatless in Seattle: AI Ate My Headcount

Some time ago I wrote that, for SaaS companies, user-based pricing was generally a bad pricing model.

It was bad because it put a brake on the adoption of your software in your customer’s organisation, whereas making it free (and using some other metric as the basis for pricing) would better encourage adoption and pervasiveness. And whilst pervasiveness is not stickiness, it is probably the next best thing.

There’s now another reason to be wary of user-based pricing.

The recent fall in the share prices of listed software companies, driven by the fear that AI will replace software (or, at least, make it really easy for anyone to produce software) has hit companies using seat-based pricing the worst.

The logic is simple. If AI increases efficiency, you need fewer people to do a job. If your customers are going to be employing fewer people, and your pricing is based on numbers of people, then you are going to be making less money.

But there’s another reason. As even user-based software moves to AI, the costs of delivering the service become increasingly unpredictable. The increasing costs of compute and data are making traditional pricing models unsustainable, pushing companies to pricing models that are usage, rather than user, based.

As one article put it: “The AI disruption is not merely a product threat; it is a fundamental reset of how enterprise software is priced and valued. The industry is accelerating toward consumption-based and hybrid pricing structures, where cost is tied to actual usage or outcomes, not to a static headcount.”

24th February 2026