Outcome-Based Pricing
Outcome-based pricing charges a customer for a result the software produced — a support ticket resolved, a debt collected, a document processed, a sale closed — rather than for a seat, a licence or a unit of usage. It has spread with AI agents, whose vendors sell work done rather than tools, and it changes what a buyer of such a company is buying: not recurring revenue, but a share of the customer's outcomes.
- Also known as
- Outcome pricing, Pay-per-outcome, Results-based pricing
- Used at
- AI customer service (Intercom's Fin charges $0.99 per resolved conversation), collections, claims processing, sales and lead qualification, back-office automation
- Binding
- yes — the contract must define the outcome, who measures it, what counts as the vendor's, and the price per unit; disputes over attribution are the norm without it
- Typical structures
- a fixed price per outcome; a percentage of value recovered or generated; a platform fee plus outcome charges; a minimum commitment with outcome pricing above it
- Typical effect
- revenue that tracks the customer's volume, so it is less recurring than a subscription and more than usage — buyers value it somewhere between
- Parties
- the vendor, the customer, and the finance teams on both sides who must agree what happened
How outcome-based pricing works in practice
The pitch is alignment: the customer pays only when the software delivers, and the vendor is paid in proportion to the value it creates, not the seats it sold. A support platform charging per resolution earns nothing on the question it could not answer; a collections agent taking 5% of what it recovers earns nothing on the debtor who never pays. It removes the customer's risk that a licence goes unused and shifts to the vendor the risk that the product does not work.
The difficulty is in the definition. What is a resolution — the customer said "thanks", or did not come back for seven days, or a human agent never touched the case? Who gets credit for a payment that arrived after the AI agent's call and a letter from the customer's own team? A contract that does not answer these produces an invoice dispute every month. Good agreements define the outcome operationally, name the system of record that counts it, and set a reconciliation process.
For the vendor's own valuation, outcome revenue is a hybrid. It repeats, because the customer's tickets and debts recur, but it moves with the customer's volume and can be cut by the customer's own improvement. Quality-of-earnings work on an outcome-priced business reads the volume trend, the concentration by customer, and the gap between outcomes claimed and outcomes paid.
Example
A collections agency licenses an AI voice agent at 5% of amounts recovered on early-arrears accounts, with a $20,000 monthly minimum. In its first full quarter the agent works 40,000 accounts and recovers $6.2m: $310,000 of fees, against the $180,000 the agency paid for the dialler seats it replaced. The vendor's revenue from this customer is $103,000 a month — but a seat licence would have been $60,000 a month whatever happened, and when the agency's book of early arrears shrinks by a third in a good year, so does the vendor's revenue. A buyer of the vendor models this customer at the subscription equivalent it displaced, with the excess treated as variable.
Common mistakes
- An outcome defined by the vendor's dashboard rather than the customer's system of record. - No attribution rule for outcomes the customer's own actions contributed to. - A vendor pricing on outcomes it cannot measure inside the customer's environment. - Investors reading outcome revenue as ARR and paying a subscription multiple for it. - No minimum commitment, so a pilot that produces no outcomes costs the vendor its whole delivery effort.