What Is Outcome-Based Pricing and Why Are AI Vendors Adopting It in 2027?
Outcome-based pricing means you pay when the software delivers a result, such as a resolved support ticket or a completed task, instead of paying for each user seat or for raw usage. AI agent vendors are adopting it because an agent does the work that a seat-holder used to do, so a per-seat price stops matching the value, and because buyers are asking to pay for results rather than for assistive tools.
The model is real and priced today. On the vendors' own pages (read October 21, 2026), Intercom's Fin lists $0.99 per outcome and Zendesk lists $1.50 per committed automated resolution. It is not risk-free: the hard parts are defining what counts as an outcome, proving the AI caused it, and forecasting a bill that moves with volume.
Below is how it works, why it is spreading, the risks, and what buyers and sellers should write into the contract.
What outcome-based pricing means
Software has usually been sold in one of two ways. Per seat means you pay for each named user whether or not they do much. Per usage means you pay for a technical meter such as messages, tokens or actions. Outcome-based pricing moves the meter to the result: you pay for each resolved ticket, completed task or qualified lead, and you do not pay when the agent fails or hands off.
- Per seat: a fixed fee per user. Cost follows headcount, not results.
- Per usage: a fee per action or conversation. Cost follows activity, whether or not it solved anything.
- Per outcome: a fee per agreed result. Cost follows what was actually accomplished.
In between sit hybrids, such as a seat fee plus a usage or resolution fee, and committed bundles where you prepay a block of outcomes at a lower rate. Most real contracts are hybrids.
Real prices from three vendors
These figures were read on each vendor's own pricing page on October 21, 2026. Vendors change prices often, so check the live page before you rely on them.
- Intercom Fin: the Fin pricing page lists $0.99 per Fin outcome. Resolutions, procedure handoffs, disqualifications and self-serve routings are $0.99 each, and qualifications are $9.99 each. You are charged for one outcome per conversation, and not charged when a conversation is simply passed to your team without an outcome. For Fin on a non-Intercom helpdesk the page states a minimum of 50 outcomes per month, and it says minimum commitments apply.
- Zendesk: the Zendesk pricing page bills automated resolutions, which it says means you pay only for customer requests that were successfully resolved. Committed automated resolutions are $1.50 each and pay-as-you-go resolutions are $2.00 each on Support Team, Suite Team and Suite Professional. Plans include 5 automated resolutions per agent per month on Support Team and Suite Team, and 10 on Suite Professional. No per-resolution price is listed for Suite Enterprise, which is sales-assisted.
- Salesforce Agentforce: the Agentforce pricing page is a useful midpoint rather than a pure outcome model. It lists $2 per conversation, and separately a Flex Credits model in which a standard Agentforce action uses 20 Flex Credits, shown as US$0.10 in the page's examples. Those are per-conversation and per-action meters, not per-result charges. It also lists an Agentforce User License at $5 per user/month, which requires Flex Credits. The page says its information is for reference only and subject to change.
How vendors define an outcome
The definition is the whole product. Intercom's help article on Fin outcomes says a resolution is counted when, after Fin's last answer, the customer either confirms the answer was satisfactory or exits the conversation without asking for more help. The first is a confirmed resolution and the second is an assumed one. The same article says you will never be charged for an outcome that did not happen, and that asking for a person is not billed and a Procedure that does not finish is not billed.
Fin's pricing page puts it more shortly: a resolution means no further help is requested after Fin's last answer. Notice what that implies. A customer who gives up and leaves silently can look the same as a customer who was helped. That gap between a proxy and a true result is where most disputes will start.
Why AI agent vendors are moving to it
There are three reasons, and the first is the vendors' and buyers' shared logic rather than a published statistic.
- The seat stops matching the value. An agent that answers customers does work that humans used to do. If it replaces some of that work, a price per human seat can shrink as the product succeeds. Charging per result keeps price tied to value delivered. This is reasoning, not a number from a report.
- Buyers want to pay for results. Gartner said on April 2, 2026 that by 2028 over half of all enterprises will stop paying for assistive intelligence such as copilots and smart advisors, and will favor platforms that commit to workflow results. It is a prediction, but it shows the direction buyers are being pushed.
- It lowers the buyer's risk of trying. If you only pay when the agent resolves something, a pilot costs little when it fails. Fin's page says to pay only when Fin delivers value. A vendor that is confident in its product can use that as a sales tool.
There is a counterweight. Gartner said on January 26, 2026 that by 2030 the cost per resolution for generative AI will exceed $3, citing rising data center costs, AI vendors pivoting from subsidized growth to profitability, and complex use cases that consume more tokens. That prediction matters for pricing: if a vendor fixes a price per resolution, a rising cost to serve squeezes the vendor, so expect per-result prices and minimums to be revisited.
The risks
- Defining the outcome. If the definition is loose, the vendor can bill for results that did not help, and the buyer can dispute results that did. The assumed-resolution case above is the classic example.
- Attribution. A ticket closed after the AI answered may have closed anyway, or a human may have done part of the work. Decide up front whether partial help counts, and how a handoff to a person is treated.
- Revenue predictability for the seller. Seat revenue is steady; outcome revenue moves with ticket volume, seasonality and model quality. That is why you see committed bundles and minimums, such as Fin's 50 outcomes per month minimum on a non-Intercom helpdesk and Zendesk's cheaper committed rate against its pay-as-you-go rate.
- Bill predictability for the buyer. If volume doubles you pay more, which is fair if the value doubled but hard to budget. Fin's page mentions configurable alerts, limits and real-time reporting, which exist for this reason.
- Gaming and quality drift. When money depends on a counted result, either side can be tempted to steer toward cheap, easy resolutions. Audit samples of billed outcomes rather than trusting the dashboard.
Practical advice for buyers
- Get the definition in writing, including whether silent exits count, what happens on a handoff, and whether a repeat contact on the same issue is billed twice.
- Sample and audit. Each month, read a random set of billed outcomes and check them against what your team would call resolved.
- Cap the bill. Ask for spend alerts and a hard limit, and model your cost at your highest ticket volume, not your average.
- Compare on cost per true resolution, not the headline price. A cheaper price with a looser definition can cost more per genuinely solved problem.
- Negotiate the minimum. Where a minimum commitment applies, check what happens to unused volume. Salesforce's page, for instance, says unused Flex Credits do not roll over into subsequent subscription terms.
- Keep an exit path. Make sure customers can always reach a person, since a human handoff is not a billable outcome on the pages above.
Practical advice for sellers
- Pick an outcome you can measure automatically, close to the work the product does and visible soon after. A result that takes months to confirm is hard to bill.
- Offer a hybrid. A small platform fee or a committed bundle protects your revenue while the per-result part aligns with value. The vendors above all use some form of commitment.
- Publish the definition. Intercom documents what counts and what does not count as a billable outcome. Clear rules reduce billing disputes.
- Watch your cost to serve. Model your own inference cost per outcome. If costs rise, as Gartner predicts for generative AI customer service, a fixed price per result can turn unprofitable.
- Give buyers controls. Usage alerts, limits and reporting build trust faster than discounts.
Questions people ask
What is outcome-based pricing in simple terms?
You pay when the software delivers an agreed result, such as a resolved support conversation, instead of paying per user seat or per unit of activity.
How much does outcome-based AI pricing cost?
It varies by vendor and definition. On their own pricing pages read October 21, 2026, Intercom Fin lists $0.99 per outcome and Zendesk lists $1.50 per committed automated resolution and $2.00 pay-as-you-go. Check the live pages, since prices change.
Is Salesforce Agentforce outcome-based?
Not purely. Its pricing page lists $2 per conversation and a Flex Credits model where a standard action uses 20 Flex Credits, which are per-conversation and per-action meters rather than per-result charges.
What is an assumed resolution?
In Intercom's definition it is a resolution counted when the customer exits the conversation without requesting further assistance after Fin's last answer, as opposed to a confirmed resolution where the customer says the answer was satisfactory.
What are the biggest risks for buyers?
A loose definition of an outcome, unclear attribution when humans and AI both work a case, and a bill that rises with volume. Get definitions in writing, audit samples of billed outcomes and set spend limits.
Will all AI software move to outcome-based pricing?
Unlikely. Gartner predicts over half of enterprises will favor platforms that commit to workflow results by 2028, but that is a forecast, and many contracts will stay hybrids of seats, usage and commitments.
Sources
- Fin AI Agent pricing (fin.ai) · Per-outcome prices, minimums and billing rules; read October 21, 2026
- Intercom Help: Fin AI Agent outcomes · How a resolution is counted and what is not billed; read October 21, 2026
- Zendesk pricing · Automated resolution prices and included allowances; read October 21, 2026
- Salesforce Agentforce pricing · Per-conversation and Flex Credits pricing; read October 21, 2026
- Gartner press release, April 2, 2026 · Enterprises favoring platforms that commit to workflow results by 2028
- Gartner press release, January 26, 2026 · GenAI cost per resolution prediction
Prices and definitions were read from the linked vendor and Gartner pages on October 21, 2026. Vendors change prices and terms often, and Gartner's figures are predictions, so confirm both before signing.
Top photo: perzon seo / CC BY 2.0 via Wikimedia Commons