Intercom Fin Pricing: What Per-Resolution Billing Actually Costs a Small Team

Fin is the strongest AI support agent on the market right now. That's not in dispute, and this article isn't going to pretend otherwise.

What's worth working out before you adopt it is what outcome-based billing does to a small team's monthly bill, because the pricing model is genuinely good for some companies and genuinely bad for others, and the deciding factor isn't your headcount. It's your ticket volume.

Fin is not standalone. You pay for seats, then you pay for Fin on top.

Seats (per seat, per month)

Fin: $0.99 per outcome, on top of seats. A minimum monthly commitment applies (Intercom's example: 50 outcomes).

Copilot (AI assistant for your human agents, a separate product): $29/agent/month annual, $35 monthly. Each agent gets 10 free Copilot conversations a month.

Intercom changed its billing unit. It used to bill per resolution: Fin handled the issue end-to-end, no human involved. As of March 2026 it bills per outcome, which Intercom defines as Fin "successfully completing the action it was configured to perform, as part of a conversation."

An outcome can be counted even when a human agent still finishes the ticket.

If Fin gathers the customer's details, takes an action, and then hands off to your team, that's a billable outcome and a human still spends time on it.

Intercom's reasoning is defensible: Fin did real work in that scenario, and "success stopped being binary" as the product got more capable. But for budgeting purposes, be clear-eyed: the billable event is now broader than it was, so per-unit cost comparisons against pre-March-2026 numbers aren't apples to apples.

Intercom counts an outcome when a customer confirms resolution, when they don't ask for more help after Fin responds, or when Fin completes a workflow.

Assume a 3-person team on Advanced ($85/seat = $255/month in seats):

At 1,000 outcomes a month, Fin alone costs roughly four times your seat spend. The line crosses early: past about 300 outcomes, the AI is the majority of your Intercom bill.

This is the structural property of per-unit pricing: your bill grows in direct proportion to your success. Fin working well and Fin being expensive are the same event. If your margins are healthy per customer, that's fine and arguably ideal. If you're pre-revenue or running thin margins on a low-priced product, a support bill that scales linearly with usage is a genuine planning problem.

Intercom states an average resolution rate across customers of 76%, and says it has risen every month. That figure is from Intercom, measured by Intercom.

Independent looks at the same question come out lower. Analyses of Intercom's own published case studies land nearer 40–50%. Linktree at 42%, Robin at 50% are the commonly cited ones. One independent test across 500 tickets at four small businesses reported a 38% average.

Both things can be true. Intercom's 76% is an average across 12,000+ customers, weighted toward companies with mature, well-maintained knowledge bases and significant configuration investment. A four-person startup with a thin help center is not that average customer.

The consistent finding across every source: resolution rate is determined far more by knowledge base quality than by the model. Reported rates across Intercom's base run from 25% to over 80%, and the spread tracks documentation quality, not which AI is under the hood.

Plan with 40–50% for a new deployment on modest docs. If you beat it, good. Budgeting on 76% from day one will overstate your savings and understate your human workload.

Straightforwardly, when:

• Your ticket volume is high enough that human time dominates the cost. Human-handled tickets benchmark around $8–$35 depending on channel and complexity. Against a $20 ticket, $0.99 is excellent economics.
• You already run Intercom. Fin is a toggle, not a migration.
• Your knowledge base is already good. You'll land near the top of that 25–80% range instead of the bottom.
• You need what Fin does well: multi-channel, deep workflow actions, mature analytics, enterprise controls.

• Low volume. Under ~100 conversations a month, the seat cost dominates and you're paying enterprise tooling prices for a trickle. The 50-outcome minimum sets a floor regardless.
• Unpredictable volume. A launch, a Product Hunt day, or an outage produces a support spike and a bill spike simultaneously.
• Website-chat only. Fin's multi-channel breadth is a large part of its price. If you only need a widget on your marketing site, you're buying capability you won't use.
• Thin margins. Linear cost scaling against a $10/month product is a squeeze.

Don't compare Fin's price against other AI tools first. Compare it against what you currently spend handling those tickets by hand.

Take a month of tickets. Multiply by a realistic per-ticket cost. $8–$16 is a defensible starting range for chat/email in SaaS, higher if the work is complex. That's your baseline. Then model Fin at a 40–50% resolution rate: that share becomes $0.99 each, the rest still costs you human time, plus seats.

If the total is lower than your baseline, Fin pays for itself. If your baseline is small because your volume is small, no AI support tool pays for itself yet, and the right move is to fix your docs and wait.