The seven models
Per seat: predictable, penalizes adoption. Tiered plans: simple, forces packaging discipline. Usage-based: aligns cost with value, unpredictable bills. Per-seat plus usage hybrid: the emerging default. Flat rate: frictionless, leaves money on the table. Outcome-based: charges per result, hard to attribute. Freemium: a distribution strategy wearing a pricing costume.
How to choose
Price against your value metric: the unit that grows when your customer gets more value. If value scales with people, seats work. If it scales with volume, meter it. If your AI product does work autonomously, seats actively undercount your value, which is why agent-era products increasingly price per task or per outcome.
The AI pressure
When software performs the work instead of assisting a human, the seat stops being the value unit. Buyers have noticed; procurement teams now routinely ask what a task-based price would look like. Have an answer even if you keep seats.
FAQ
How often should we raise prices?
Review annually, change when your value delivery has demonstrably grown. Grandfather existing customers for a defined period, not forever.
Should early-stage startups discount?
Trade discounts for something: annual prepay, a case study, a reference call. Naked discounts anchor the account permanently low.
Is freemium dead?
No, but it is a marketing expense. Run it only if free users convert, refer or generate data that compounds your product.
