Key Takeaways

• HubSpot charges $0.50 per resolved conversation for Breeze Customer Agent, replacing $1.00 per conversation in April 2026.

• Intercom's Fin charges $0.99 per outcome and bills once per conversation, however many questions it answers.

• A result is billable only when a system detects it, timestamps it, and attributes it to one customer.

• Outcome-based pricing shifts delivery risk to the vendor, so failed attempts drive margin as much as wins.

• AI agent products drive most adoption, because an agent's work ends in a logged result seats describe badly.

What counts as an outcome you can bill for?

An outcome qualifies when it's defined in writing, emitted as an event the moment it occurs, and owned by one party. Anything softer becomes an invoice dispute. Published rates show the pattern:

Intercom publishes the tighter specification. Fin counts an outcome when the customer confirms the issue is resolved, when the customer stops asking after Fin replies, or when Fin finishes a workflow. It bills once per conversation, which caps what a long thread costs.

HubSpot quotes its rate without publishing the resolution criteria, the common gap. Defining "resolved" is harder than setting the number, and our walkthrough of outcome-based pricing covers how teams write that definition.

The unit also has to be countable in a stream you already collect, which puts outcome models closer to consumption-based pricing than to a subscription. Most AI agent pricing launches reuse that pipeline and move the billable trigger down it.

Where does outcome-based pricing work and where does it fail?

It works where the vendor controls the result and the result leaves a clean record. It fails where either half is missing:

• Works: ticket resolution, lead qualification, collections recovered, and other tasks an agent finishes alone.

• Works: categories with a clear counterfactual, where the buyer knows what a human resolution costs.

• Fails: outcomes the customer influences, such as revenue lifted or churn avoided, where attribution becomes an argument.

• Fails: long horizons. If the outcome lands 90 days after the work, collection drifts away from cost.

• Fails: costly failed attempts. Every unresolved conversation burns inference spend with nothing to invoice.

Forecasting also gets harder. Once the billable count tracks product performance instead of what the customer buys, finance needs a floor or a commitment.

Related terms

Related entries worth reading:

• Consumption-Based Pricing prices units burned, not results.

• AI Agent Pricing covers every structure used to charge for autonomous agents.

• Hybrid Pricing Model pairs a platform fee with an outcome charge, how most outcome deals ship.

• Credit-Based Pricing sells prepaid credits an outcome charge can draw down.

FAQ

How is outcome-based pricing different from usage-based pricing?

Usage-based pricing bills the work performed, outcome-based pricing bills only the work that succeeded. An agent handling 1,000 conversations and resolving 650 bills 1,000 units under usage pricing and 650 under an outcome model.

Why are AI companies adopting outcome-based pricing?

AI agents produce discrete, loggable results, so the outcome is easy to count. Token prices fall as models get cheaper, and seats don't hold up when software does the task instead of a person. Pricing the result keeps revenue tied to what the buyer is buying.

Can you mix outcome pricing with a subscription?

Yes, and most vendors do. A platform fee covers access, and the outcome charge sits on top, often drawing against a prepaid balance. The fixed fee gives finance a revenue floor while the outcome charge captures the upside.