Outcome-based pricing means a company pays its AI vendor per agreed result, such as a customer request resolved without a human agent. It differs from per-minute, per-call, and subscription pricing, which charge for usage or access.
The idea is simple. Most of the commercial detail sits in the definition of an outcome.
How outcome-based pricing works
Definition. Buyer and vendor agree in the contract what counts as an outcome and how it is detected.
Price. Each outcome has a unit price, sometimes with volume tiers, a minimum commitment, or a cap.
Measurement. The vendor's platform counts outcomes. The buyer checks the count against its own records, such as transfers, repeat contacts, and CRM cases.
Invoice. The buyer pays for the outcomes delivered in the period. Conversations that fell short of the outcome are free or billed at a lower rate.
What "outcome" can mean
Resolution. The customer's request was handled without a transfer to a person. Vendors differ on the evidence: a customer's confirmation, a period of silence after the last message, or an automated review of the conversation.
Containment. The contact never reached a human agent. This is the loosest definition, because a customer who gives up also counts.
Task completed. A specific action was carried out, such as a booked appointment, a changed address, or a registered claim.
Business result. A cancellation was prevented, a product was sold, or a payment was collected.
Public definitions show the range. Sierra lists outcomes such as a resolved support conversation, a saved cancellation, an upsell, or a cross-sell. It says escalated cases are in most cases free. Zendesk bills per automated resolution and evaluates only conversations that were not escalated. As of November 2026, Intercom lists its Fin agent at $0.99 per outcome.
Four points are worth settling in any definition. The first is a customer who returns with the same problem a day later. The second is an abandoned conversation. The third is a transfer with all details already collected. The fourth is the buyer's right to audit the count.
Outcome-based pricing vs per-minute, per-call, and subscription pricing
Each model puts a different risk on each side of the contract.
Per-minute pricing. Common for voice. The bill follows talk time, so it grows with long calls and includes calls that end in a transfer. Minutes are easy to verify. The buyer carries the performance risk.
Per-call or per-conversation pricing. A flat fee for each interaction, whatever the result. The buyer carries the risk that calls go unresolved. The vendor carries the risk of long conversations.
Subscription pricing. A fixed fee for a platform, a number of seats, a volume tier, or a number of concurrent sessions. Budgets are predictable. The buyer pays the same amount when volume or automation is low, and the vendor carries the risk of heavy use within the tier.
Outcome-based pricing. The vendor is paid when the result is delivered and carries the performance risk. The buyer carries two others: a bill that rises with volume and success, and a definition that may count easy cases generously.
Contracts often blend the models, for example a platform fee plus a price per resolution.
To compare offers, it helps to convert each one to a cost per resolved contact. For a per-minute voice offer, that is the average minutes per call, multiplied by the price per minute, divided by the share of calls resolved. Costs outside the unit price belong in the comparison too, such as telephony, implementation, and integration work.
When outcome-based pricing matters for a contact center
The model comes up when a business case goes to finance and procurement. A price per resolved contact can be set directly against the current cost of a contact handled by a person (see ROI of AI).
Telecom. An outage can produce a surge of near-identical contacts within an hour. It is worth agreeing in advance whether each one counts as an outcome, and whether a cap applies.
Banking. The definition usually needs a compliance condition. An outcome is then a request that was resolved with the identity check passed and the required disclosures given.
Insurance. A useful outcome is often a completed step, such as a first notice of loss with every required field. The claim decision stays with the insurer.
Frequently asked questions
What counts as an outcome in AI customer service?
The contract decides. Common choices are a resolved conversation, a completed task such as a booking, or a business result such as a saved cancellation. A clear definition names the evidence, the time window for repeat contacts, and the treatment of transfers and abandoned conversations.
How are AI voice agents usually priced?
Voice agents are most often quoted per minute, per call, per resolved call, as a subscription, or by concurrent sessions. Many offers combine a fixed platform fee with a usage or outcome price.
Is outcome-based pricing cheaper than per-minute pricing?
It depends on call length and on the share of calls resolved. Short calls with a high resolution share can cost less on a per-minute plan. Long calls, or a low resolution share, tend to favor paying per outcome. Converting both offers to a cost per resolved call shows which applies to your call mix.
What are the risks of outcome-based pricing for the buyer?
There are three main ones. The bill is harder to forecast, because it rises with volume and with success. The definition may count easy requests, or customers who gave up, as outcomes. Finally, the vendor measures the result it is paid for, so audit rights matter.
How we approach this at Zowie
Whatever the pricing unit, the count has to be provable. At Zowie, the language model speaks. A separate, deterministic Decision Engine returns every amount, rate, offer, disclosure, and escalation from your rules, with an audit log. To talk through your call types and volumes, you are welcome to book a demo.
Explore: ROI of AI, Build vs buy, Deflection rate, Automated resolution rate, Voicebot, Contact center AI