Fin's $0.99 per resolution, decoded

Outcome-based pricing sells well: you only pay when it works. In practice, the definition of "works" belongs to the vendor, and the gap between sticker price and actual invoice comes down to details that aren't in the brochure. Here's the line-by-line read.

What gets billed

The official documentation lists two billable outcomes at $0.99 each.

Two points deserve emphasis. First, billing happens once per conversation, however many questions are asked inside it. Second, a resolution requires Fin to have actually answered: an exchange of pleasantries doesn't count, and isn't billed.

What isn't billed, and nobody knows

This is the useful part, because it's counter-intuitive and it changes how you read an invoice.

• Escalations triggered by your workspace rules or by Fin's own frustration detection.
• Procedure failures and technical errors.
• Abandoned conversations, meaning the customer doesn't answer clarifying questions.
• Failed handoffs.

A wrong answer costs twice

Here's the mechanism outcome pricing makes invisible. Fin answers beside the point. The customer doesn't push back in the thread, they leave. The conversation closes without a further request for help: that's a resolution, billed at $0.99.

Two days later the same customer comes back. Not in the same conversation, in a new one. Fin answers again, the customer finally gets their answer or moves to a human. Since billing is per conversation, you pay a second time.

A wrong answer therefore doesn't cost $0.99, it costs $1.98 at best, plus human time if it escalates, plus the satisfaction hit. And since the native reopen counter only sees reopened conversations, not new threads, that double payment shows up nowhere in your reports. We cover that blind spot in our piece on reopen rate.

The sticker isn't the invoice

Outcomes are one line. Here are the others, at publicly observed 2026 rates.

Full seats include Copilot up to ten conversations a month, beyond that it's the add-on. On a non-Intercom helpdesk the $49 plan sets a floor of 50 monthly outcomes; on an Intercom plan there is no floor.

A worked example

Ten agents on the Advanced plan, Copilot for everyone, around 1,500 Fin resolutions a month. Seats alone come to $10,200 a year. The real invoice lands near $31,500, roughly triple. The market-wide pattern is actual spend running 60 to 80% above seat cost.

The three levers that work

Shrink scope before volume. On topics where Fin often fails, each attempt generates a billed resolution then a billed catch-up conversation. Pulling a topic out of Fin's scope lowers the invoice more reliably than any prompt tuning.

Treat the knowledge base as a budget line. A missing article on a high-volume topic gets paid for in double billings every month. The return on half a day of writing is computable, and it usually beats the return on negotiating with sales.

Measure quality independently. The vendor's resolution rate can't tell a good answer from an off-target one the customer gave up on. Both bill the same. Without an external measure, you have no way of knowing what share of your invoice pays for wrong answers.

The optimisation trap

One reflex to avoid: driving the outcome count down to drive the invoice down. Fewer Fin resolutions means more human conversations, and a human conversation costs far more than a dollar. The goal isn't fewer outcomes, it's a smaller share of outcomes that resolve nothing.

That's precisely what ResponZ quality scoring makes visible: every Fin answer scored, and the amount represented by the answers that don't hold up.

The rates quoted here are those publicly observed in 2026. The outcome model is still in place, but roadmap and packaging now sit with Salesforce: we cover what that changes in our take on the acquisition.