We are launching an AI product. How do we choose between charging per seat, per use and per outcome?
Start from who or what does the work. If people use the product to do their own job better, a seat still makes sense and buyers know how to budget for it. If the product does the work itself, charge for the work: per use or per credit where usage is easy to count, and per outcome (a resolved ticket, a booked meeting, a reconciled invoice) only where you and the customer can agree what counts and your product measures it rather than leaving it to argument. Most teams land on a hybrid: a platform fee that includes an allowance of credits, with overage above it, which gives the buyer a predictable number and covers your model costs on heavy users. Set the price after a month of metering cost per customer, because a small share of users usually drives most of the inference bill, and plan for gross margins nearer 50 to 65 percent than the 80 percent of classic SaaS. If you sell to Indian SMBs, show the allowance as a fixed monthly plan in rupees, since a meter reads as an open-ended bill. Expect to reprice within a year, and write your contracts so that you can.
Go deeper
5 resources, 5 link-checked.
📰 Newsletter
✓ Link checkedFreeAdvanced
The clearest statement of why per-seat pricing breaks when software does the work, with the support agent example (115 dollars per seat per month becomes a price per resolved ticket) and the variable cost problem underneath AI products.
The margin defence most companies are actually reaching for. Poyar's number is the one to remember: roughly 70 to 80 percent of token consumption comes from about 10 percent of users, which is why a flat seat price on an AI product bleeds quietly.
The pricing advisor with the widest view of AI deals on how to charge when every use has a real cost: hybrid models, floors, and pricing the outcome without giving away margin. Directly addresses the gross margin trap.
Survey data from 230 B2B software and AI companies, so you can see what your peers are actually charging rather than what conference talks claim. Notable findings: hybrid pricing is now dominant and AI products carry thinner margins than classic SaaS.
Gives you the current benchmarks to set expectations with your board (roughly 45 to 53 percent depending on how much of the stack you own) and four concrete moves: pick a margin posture deliberately, meter cost per customer, route to cheaper models by default, price with usage.
