The Seat Is Dead: Pricing for Work Humans No Longer Do
Per-seat pricing was always a proxy for value scaling with human headcount. When agents do the work the seats used to do, you bill the customer for empty…
Per-seat pricing was a proxy. It assumed that value scaled with the number of humans logged in doing the work, so you billed for the humans. Agents break the proxy. When the software does the work that the seats used to do, seat-based pricing charges the customer for empty chairs while the vendor captures almost none of the value it actually created. The pricing model and the value model have come apart, and no discount schedule closes the gap. Pricing is not a packaging exercise you run once a year. Architecture sets pricing, and the architecture shifted underneath every company still selling by the seat. The per-seat model was never really about seats. It was a measurement convenience. Value in software was hard to meter directly, but it correlated with how many people used the tool to do their jobs, so vendors billed per person and called it close enough. The seat was a proxy for human labor applied through the software. As long as humans were the ones doing the work, the proxy tracked the value well enough that nobody questioned it. Every proxy holds only as long as the thing it stands in for stays constant. The seat held because the relationship between headcount and value held. The moment that relationship breaks, the proxy stops measuring anything real, and you are billing on a number that no longer corresponds to what the customer receives. An agent does the work without occupying a seat. The customer gets the output, the contracts processed, the tickets resolved, the analysis produced, with fewer humans logged in, not more. Under per-seat pricing, that is a catastrophe disguised as efficiency: the more of the work your software automates, the fewer seats the customer needs, and the less you earn for delivering more value. You have built a pricing model that punishes you for being good at the thing you sell. The value you create goes up and the revenue you capture goes down, because the two are pegged to opposite trends. Discounting cannot fix this, and neither can a new pricing tier, because the failure is in the denominator. Seats are the wrong unit once seats stop tracking work. You can adjust the price per seat all you like; you are still measuring the wrong thing. The reflex is to treat pricing as a marketing decision, revisited annually, tuned for conversion. That framing is what leaves companies exposed. The unit you price on is a structural decision about whether your revenue tracks the value you create. Bill on usage and your revenue rises as the customer does more, which is exactly when they value you most. Bill on outcomes and your revenue rises as you deliver results, aligning you with the customer by construction. Bill on seats in a world of agents and your revenue falls as your product improves. These are not packaging variants. They are different architectures with different futures. Choosing the denominator is the work. It requires understanding what the customer actually values, how to meter it credibly, and how to build the commercial system, the contracts, the forecasting, the data layer, that can run on a non-seat unit. That last part is why many companies cling to seats: seats are easy to count and easy to forecast. Usage and outcomes require a real metering architecture underneath. The companies that build it capture the value of the automation. The companies that do not watch it accrue to the customer while their per-seat revenue erodes. We rebuild commercial models when the pricing unit and the value have come apart: choosing the new denominator, building the metering layer, and re-architecting the forecasting and contracts around it. The shift off seats is not a price change. It is a system change.