What it is
Outcome-based pricing means paying for the result an agent achieves rather than for a subscription or seats. It closed a ticket, you pay for the resolution. It booked a meeting, you pay for the meeting. The price is tied to value, not to access.
Where it came from
SaaS lived on per-seat pricing for decades. But an AI agent replaces work rather than handing a human a tool — so the old model broke. Intercom was the first to loudly switch to per-resolution pricing for Fin, then evolved it into per-outcome.
Why it took off
Intercom Fin: $0.99 per outcome, growth from $1M to $100M+ ARR on this model, up to 80%+ of support volume, around 1M resolutions a week — with a money-back guarantee if the resolution target is missed. The buyer pays only for a real result, which removes the fear of "we bought it and it never took off".
How to use it today
- If you sell an AI product, consider pricing per result: it is a strong answer to buyer risk.
- Define clearly what counts as an "outcome" (and do not let the system farm the easy ones).
- As a buyer, prefer per-outcome for agentic products: you are paying for value.
- Build in margin: if the token cost per result exceeds the price, the model loses money.
What to watch out for
Defining success is not trivial: Intercom itself moved away from a binary "resolution" to an "outcome" because the agent started working alongside a human. The risk is arguing over whether something counts as a result. You need transparent metrics and guarantees.