Outcome-based pricing
Outcome-based pricing is a pricing model that charges customers for a completed business result — a resolved support ticket, a qualified lead, a completed task — rather than for tokens, API calls, or seats, so price tracks the value delivered rather than the resources consumed to produce it.
How outcome-based pricing differs from usage-based and seat-based pricing
Usage-based billing charges per unit of consumption — a token, an API call — and seat-based billing charges a flat fee per user regardless of how much of a feature that user actually consumes. Outcome-based pricing charges for neither: it prices the business result itself — a resolved support ticket, a qualified lead, a completed task — so the customer pays only when the AI feature actually delivers the outcome it was built to produce, independent of how many model calls, tokens, or agent steps that outcome took to produce.
That shift moves the pricing conversation from how much AI was used to how much value was delivered, which is easier for a buyer to justify internally and easier for a seller to sell against the known cost of an existing manual process.
Why AI makes outcome-based pricing newly viable
Outcome-based pricing has always been possible in principle — a services contract can be billed per deliverable — but it has historically been hard to price into a software product because a compute or seat allotment does not map cleanly to a business outcome. AI changes that because outcomes well-defined enough for a model to attempt — ticket resolution, lead qualification, task completion — are also outcomes a system can detect and measure automatically at the moment they happen, using the same event stream that already feeds cost attribution and usage metering. Once an outcome is both AI-produced and machine-verifiable, pricing on it directly becomes a metering problem rather than a manual-audit problem.
Worked example: price per outcome vs. cost per outcome
A support-automation product charges $2.00 per resolved ticket. Each resolution runs a chain of model calls — retrieval, reasoning, and response generation — each billed at $3 per 1,000,000 input tokens and $15 per 1,000,000 output tokens. A typical ticket resolves in 3 calls averaging 800 input tokens and 300 output tokens per call.
Margin on a typical resolved ticket
- Cost per call = (800 tokens × $3 ÷ 1,000,000) + (300 tokens × $15 ÷ 1,000,000) = $0.0024 + $0.0045 = $0.0069
- Cost for a typical 3-call resolution = 3 × $0.0069 = $0.0207
- Margin on a typical ticket = $2.00 − $0.0207 = $1.9793
The margin risk: outcome cost is variable, price is fixed
The $2.00 price per ticket is fixed, but the number of model calls needed to reach that outcome is not. A hard ticket that requires a long agent loop before it resolves might take 300 calls instead of 3.
Cost of a hard ticket at the same fixed price
- Cost for a 300-call resolution = 300 × $0.0069 = $2.07
- Net result on that ticket = $2.00 − $2.07 = −$0.07 (a loss)
- The average across many tickets can still be profitable, but any outcome whose call count runs long enough turns the fixed price into a loss — the same variable-cost-vs-fixed-price mismatch that drives margin leakage, and it needs the same cost-per-inference visibility to catch before it compounds across a whole book of customers.
Related terms
Frequently asked questions
How is outcome-based pricing different from usage-based billing?
Usage-based billing prices the resource consumed — tokens, API calls, compute time — regardless of whether that consumption produced a useful result. Outcome-based pricing prices only the completed result itself, so a task that takes ten times the inference cost to complete is billed the same as one that takes a tenth as much, as long as both actually deliver the outcome.