| | |

Intercom Fin Is Moving AI Support From Resolutions to Outcomes

ByWorkTechJournal Editorial Team

Intercom published a quiet but significant update to how its Fin AI agent is priced and measured. The shift — from resolutions to outcomes — is not just a billing change. It reflects how AI support systems are evolving from simple deflection bots into something closer to operational support infrastructure. For teams using Fin, or evaluating it, the implications run deeper than the pricing page.

What Intercom Changed

In March 2026, Intercom published a post explaining a change in how Fin’s value — and its cost — is measured. Historically, Intercom says, Fin was evaluated on whether it fully resolved a conversation on its own, without involving a human agent. A resolution meant the bot handled it end-to-end. Anything else did not count.

Intercom says it is now moving to an outcome-based model. An outcome, as Intercom defines it, occurs when Fin successfully completes the action it was configured to perform as part of a conversation. A resolution is still one type of outcome. But a new type — called a Procedure — has been added, and it works differently.

The pricing is $0.99 per outcome, and customers are charged only once per conversation, even if Fin answers multiple questions in the same session.

Why Outcomes Are Different From Resolutions

The resolution model was clean and easy to understand: did the AI close the ticket, yes or no? That worked when Fin handled simple, single-step queries. Intercom says its average resolution rate across customers is now 67%, and climbing roughly 1% per month — but also notes that Fin is increasingly handling more complex queries.

Complex cases do not always end in clean AI-only resolutions. Intercom says complex support scenarios can require Fin to gather context, read and write to external systems, execute actions, and hand off to a human agent for confirmation. A billing dispute, a subscription change, or a transaction problem might require several steps before a human needs to confirm or complete the action. Under the old model, that kind of interaction produced no billable resolution — even if Fin did most of the work correctly.

Under outcome-based billing, that handoff — if it follows a configured Procedure and reaches the designed endpoint — is billable. According to Intercom’s pricing page, an outcome is counted when a customer confirms their issue is resolved, when they stop asking for help after Fin responds, or when Fin completes a workflow or Procedure including a configured handoff.

What Procedures Reveal About the Future of AI Support

Procedures are the more technically interesting part of this update. Intercom describes them as multi-step workflows where Fin gathers context, takes action, and hands the conversation off when it is configured to do so. The key phrase is “when configured.” Fin does not autonomously decide when to hand off — the team defines when a Procedure endpoint is reached and what triggers a handoff.

According to Intercom’s Procedures FAQ, billing for Fin Procedures started on March 12, 2026. An outcome is charged when a conversation reaches either a Resolution or a configured Procedure handoff. Teams are not charged if a Procedure fails, exits without reaching a billable endpoint, or if a customer requests escalation outside the configured handoff logic.

That last point matters. If a customer asks to speak to a human outside the flow Fin was designed for, it does not trigger a charge. This protects teams from being billed for escalations they did not intend to count as completed outcomes. But it also means escalation rules need to be thought through carefully — gaps in configuration can lead to either unintended charges or missed handoffs.

Why Reporting, Guidance, and Escalation Rules Matter More Now

Outcome-based billing shifts more responsibility onto how teams configure and monitor Fin. Intercom’s reporting docs say teams can track Fin AI Agent metrics including deflection rate, resolution rate, CX score, involvement rate, escalation rate, pending rate, and content performance by individual help article or piece of content.

That reporting infrastructure matters because outcome billing is only predictable if teams understand their conversation patterns. A team with poorly scoped Procedures, weak escalation logic, or outdated support content will see billing that is harder to forecast — and harder to attribute to specific configuration problems without clean reporting.

Intercom’s Guidance documentation also describes how teams can shape how Fin responds: tone, follow-up behavior, escalation logic, and which content it prioritizes. Getting these settings right is not optional in an outcome-based model — they directly affect what gets billed and what does not. Intercom notes one hard limit: Guidance cannot influence what Fin does not already know, because Guidance rules are evaluated before Fin searches its content sources. That means content gaps are not fixable through Guidance alone.

Risks, Limits, and What Small Teams Should Watch

The outcome model is defensible in principle — you pay when Fin completes something useful, not just when it deflects traffic. But for smaller teams, a few practical concerns apply.

Cost forecasting becomes harder. Resolutions were relatively easy to estimate from historical ticket volume. Outcomes — especially as Procedures are added — require understanding how many conversations will reach billable endpoints versus exit early. Teams with volatile support volumes or complex workflows will need to monitor this carefully.

Support content quality matters more than before. If Fin answers incorrectly because of outdated or inaccurate help content, it may still register an outcome if the customer does not follow up. That is a billing risk and a customer experience risk at the same time. Accurate, current support documentation is not a nice-to-have — it is part of the cost structure.

Escalation rules need to be explicit. Under a Procedure model, when Fin hands off — and when it does not — needs to be defined deliberately. Leaving escalation behavior to defaults or guidance-level instructions, rather than explicit Procedure endpoints, affects both billing accuracy and customer experience.

Human oversight is still required. Intercom describes Fin as gathering context, executing actions, and handing off when configured — not as operating fully autonomously. Teams that assume Fin can handle complex operational workflows without careful design and ongoing review are likely to encounter problems. The “Fin replaced our support team” framing is not accurate and not what Intercom is describing.

Related Guides

Bottom Line

Intercom’s shift from resolutions to outcomes is a reasonable evolution for a support AI that is being used for increasingly complex work. The outcome model aligns billing more closely with actual value delivered — including multi-step Procedures with human handoffs. But it also means that configuration quality, escalation logic, support content accuracy, and ongoing reporting are now directly tied to both what teams pay and what their customers experience. For support leads and operations teams evaluating Fin, the question is not just “what is the resolution rate?” — it is whether the team has the infrastructure to configure, monitor, and adjust an outcome-based AI system responsibly.

Sources: Intercom Blog, Intercom Pricing, and Intercom Help Center, March–April 2026.

Similar Posts