Fin's $0.99 'Resolution' Counts the Customer Who Gave Up

Fin, an AI customer-support agent used by companies to handle inbound tickets, charges $0.99 per "resolution." A "resolution" is logged whenever a customer does not reply within 24 hours of the bot's last message, whether the customer was satisfied, gave up, or simply moved on to a competitor.

The vendor that built this definition is Fin. The price list lives on Fin's pricing page, and the rules for what counts as a billable outcome are spelled out in Fin's help article on pricing outcomes. The model is sold as outcome-based pricing: instead of paying for seats or per-message API calls, a buyer pays per closed conversation. Fin's minimum is 50 outcomes per month, and only one charge applies per conversation, no matter how many turns the bot takes.

The same price sheet bills the same shape of conversation very differently depending on the label Fin applies. A "qualified lead," Fin's term for a conversation that ends with a sales-ready handoff, costs $9.99. A "disqualified lead," the same ending the bot decided was not sales-ready, costs $0.99. A "procedure handoff," where the bot completes a client-defined script and hands off to a human, also costs $0.99. A "resolution" costs $0.99. The work the bot does is similar across all four. The label the conversation ends with is the price lever.

INLEVEL9's Issue #250 on AI cost-per-task flags the resolution category as the most elastic of the four. Fin's own pricing outcomes article defines a resolution as either "confirmed" (the customer explicitly agrees the issue is handled) or "assumed," which the help docs describe as a customer who simply does not return within 24 hours of the bot's last message. Pure greetings and clarifying questions the customer never answers are explicitly not billable. Everything else that ends in silence is.

An "assumed resolution" cannot be distinguished, from the outside, from a customer who gave up, opened a competitor's tab, decided the question was not worth answering, or finished the task themselves and walked away. The conversation looks the same in the log: a bot message, then no human reply. There is no observable signal of satisfaction. The billable unit and the underlying business outcome the buyer is paying for can diverge. The vendor is being paid for the customer stopping, not for the customer's problem being solved.

Fin builds a safety valve into the contract. If the customer reopens the same conversation after a resolution is logged, that resolution is credited back, even if the billing period has already rolled over, according to the pricing outcomes article. That is a real concession. It gives the buyer a behavioral test for whether the assumed resolution was actually a resolution. The credit-back window is a back-door satisfaction signal that the 24-hour rule does not provide on its own.

The same construct shows up across outcome-priced AI. Any vendor that sells per-outcome billing has to define the outcome, and the definition is the product. Fin publishes its definition on its own site rather than in a sales deck, which makes the case unusually clean. Customer-support agents other than Fin, sales agents priced per qualified meeting, and AI analysts priced per insight have their own versions of the gap between what the vendor counts and what the buyer would count. The portable question is the same in every case: what does the vendor log as the outcome, and is that the same thing the buyer would log if they kept the books themselves?

The price gap is where the disclosure lands hardest. A conversation Fin labels a "qualified lead" costs the buyer more than ten times as much as the same conversation labeled a "resolution" or a "disqualified lead." The bot's work is roughly constant. The label is the variable. That is what the price sheet is showing, and it is the line worth asking about the next time an AI vendor quotes per-outcome pricing.