Intercom charges roughly $0.99 per resolved ticket. Sierra charges per resolution too. The model is clean, the buyer understands it instantly, and it is the direction the market is moving.
It is also the model most likely to hurt a company that adopts it too early, and the reason is worth understanding before copying it.
What you are actually selling
Software pricing sells access to a capability and leaves the outcome to the buyer. Completed-work pricing sells the outcome and absorbs the risk of not achieving it.
That is a real transfer, not a framing change. If your agent resolves 60% of tickets when the customer expected 85%, you earn 60% of the revenue you modelled while incurring close to 100% of the cost. The customer is not unhappy in that scenario, which is the confusing part. They paid for what they got. You simply built a business on an accuracy number you had not yet earned the right to assume.
Why the floor exists
“If the outcome is recorded by a system neither of you controls, outcome pricing is viable. If it requires judgement, keep the outcome as the story rather than the invoice.”
The recommended structure from people operating these businesses is a small per-seat or platform floor plus a usage layer, held until accuracy is proven in production.
The floor is not a way of extracting more money. It covers cost to serve during the period when your accuracy is still an estimate. Cost to serve for an agent product is real and mostly variable, which is different from traditional software where the marginal customer costs almost nothing.
Once accuracy is boring, the floor can shrink toward zero and the outcome component can carry more weight. That sequence is much safer than starting at the end of it.
If the outcome is recorded by a system neither of you controls, outcome pricing is viable.
The measurement problem nobody mentions
Turn what you know into what you own.
Vibepreneur builds structured ventures from professional expertise, with positioning, launch assets, and growth systems included.
Join the WaitlistOutcome pricing requires both parties to agree on what happened. For a resolved support ticket that is relatively clean, because the resolution is a system event with a timestamp.
For most other units it is considerably murkier. Was the contract reviewed if a human then changed two clauses? Was the claim validated if it was later appealed? Was the lead qualified if it never converted? Each of those has a defensible answer and a defensible counter-argument, and you will be having that argument with your own customer every month.
The practical filter: if the outcome is recorded by a system neither of you controls, outcome pricing is viable. If it requires judgement, price on something else and keep the outcome as the story rather than the invoice.
What to do instead early on
Price on a band tied to the size of the thing your product acts on, with a usage component. Tell the outcome story in the sales conversation, quantify it in the proposal, and anchor the price against it. But bill on something that cannot be disputed.
You keep the persuasive power of the outcome without betting the company on an accuracy figure you have not yet proven.
See why outcome pricing usually is not right for the longer argument, and vertical agents eating horizontal SaaS for where the pressure is coming from.