Intercom Fin Pricing (2026): The Per-Outcome Math, Worked Through

Michael Aubry
Article Image

Intercom's pitch for Fin is one number: $0.99 per outcome, pay only when it delivers value. It is the cleanest pricing story in AI support, and it is genuinely better aligned than per-seat or per-message billing. It is also not the whole bill. This is the full math from their live pricing page in August 2026.

Disclosure: we build Murphy, an open source alternative, so read the comparison knowing that. Fin's numbers are their published prices, and the section on when Fin is worth it is sincere.

What you actually pay

$0.99 per outcome, 50 outcomes per month minimum. So the floor is about $50 a month even in a quiet month.

An outcome is not always a resolution. Their own FAQ and pricing page define it as: Fin resolves the issue end to end, or "successfully executes a Procedure you've configured to end in a handoff to a human or a workflow." Read that twice: a conversation that ends with a human finishing the work can still bill $0.99, if the handoff went through a Procedure you set up. The better you configure Fin, the more billable events you create.

The helpdesk is separate. Fin works with your existing helpdesk (Zendesk, Salesforce, HubSpot) at the per-outcome rate, or with Intercom's own helpdesk at $19 per seat per month on top.

The good parts are add-ons. The Pro add-on is "from $99/month" and it is where the genuinely impressive tooling lives: Operator (the agent that tunes your agent), CX Score, Monitors, custom AI scorecards, AI recommendations. Copilot, the assistant for your human agents, is another $35 per user per month. Voice is priced separately again.

What a real workload costs

Take a support load of 3,000 conversations a month with Fin resolving 60%:

  • 1,800 outcomes × $0.99 = $1,782/month
  • Plus 3 Intercom seats ($57) or your existing helpdesk's own bill
  • Plus Pro if you want the quality tooling ($99+)
  • Realistic total: $1,900 to $2,500 a month, scaling linearly with your growth

Our own reference point: one of our products took roughly 700 support conversations in 36 hours during a traffic spike, with an 84% resolution rate. At Fin's meter that day and a half alone is around $580, and a month at that pace lands north of $10,000. That was a spike, not a steady state, but that is exactly the point of per-outcome pricing: your best weeks are your most expensive.

The startup escape hatch is real: their Early Stage program gives eligible startups 93% off Intercom and a year of Fin free. If you qualify, that changes the math completely for a year. Then the meter starts.

When Fin is worth every cent

Honesty section. Fin is the most capable managed agent on the market and some of it has no equal:

  • Procedures with versioning, simulations, and deterministic controls: the most mature workflow system in AI support
  • Channel breadth: voice, WhatsApp, SMS, Slack, Instagram, Messenger, email
  • Works with your existing helpdesk, no migration
  • Operator and Monitors: real quality infrastructure, not dashboards for show

If you have enterprise volume, a support team whose seats are already budgeted, and the outcome rate beats what those humans cost, Fin earns the bill. The ROI calculator on their pricing page is not wrong for that customer.

The other way to do the math

Murphy is our product, stated plainly: open source, self-host free forever with no meter of any kind, or $29 a month hosted for 2,000 AI messages. The quality tooling that is a $99 add-on at Fin (conversation scoring, review queues, simulations against real past questions) is included, because on our side it is how the product improves, not an upsell. And the loop Fin does not have: real bugs become GitHub issues your coding agent fixes, and the merged PR closes the ticket.

What Murphy does not have that Fin does: voice, WhatsApp and the Meta channels, the helpdesk integrations, and years of enterprise hardening. If those are requirements, you already know which column you are in.

The structural question to ask about any per-outcome meter: do you want your support bill to be a percentage of your success, forever? For a funded team with enterprise margins, maybe yes. For a bootstrapped product where a good launch triples your conversations overnight, a meter that triples with it is the wrong shape. Murphy is on GitHub, free to self-host, and your best week costs the same as your worst.

Anything that can go wrong, will

Murphy answers your customers from your docs, hands off to humans when unsure, and turns bugs into pull requests. Free to start, open source forever.