Intercom Fin popularized per-resolution pricing for AI support. On paper it's elegant — pay for outcomes, not seats. In practice, "resolution" is defined by the vendor, minimum seat tiers sit underneath, and the bill at 3,000 conversations per month often lands north of $3,500. Here's how Fin's math actually works, and where per-response pricing comes out cheaper.
Intercom Fin is the AI agent bolted onto Intercom's broader customer support platform. Its headline pricing is per resolution — the AI answers a question, the conversation counts as resolved, you pay roughly $0.99 for that resolution. No resolution, no charge.
As positioning, it's brilliant: "pay for outcomes, not seats." As a budgeting tool, it's a little slipperier. Three things that don't usually make it into the sales deck:
None of this makes Fin a bad product. It does mean the sticker price and the actual bill are two different numbers.
Here's a realistic SMB scenario. You run an online service business. Your website gets 3,000 qualified conversations per month — people asking about pricing, availability, hours, whether you serve their zip code. Roughly 70% are the kind of thing an AI can genuinely answer; the rest escalate.
Fin's math:
For a one-admin shop, that's ~$3,100/mo all-in. Add a 3-person team and you're at ~$3,300/mo. Voice pushes it higher.
Kazozo's math on the same volume, using the Pro plan:
At 3,000 conversations a month, Fin still lands ~50% more expensive than Kazozo on a like-for-like workload, and that's before you layer on Intercom seat costs for a real team. If your conversations average 2 turns instead of 3, the included 6,000 responses cover the volume cleanly — no overages at all, and the Kazozo bill drops to $1,329/mo with voice.
The trap with per-resolution pricing isn't the headline rate. It's the definitions sitting underneath the rate. Four traps buyers most often hit:
Per AI response pricing — covered in more depth here — sidesteps most of these because a response is a discrete, observable event. Either the AI replied with text or it didn't. There's no silent-exit-counted-as-resolution loophole, no definition to negotiate.
Being fair: if you already live inside Intercom — your team uses it as the shared inbox, your reports are wired up in its analytics, your messaging tours run through its platform — Fin is a natural extension. The per-resolution model makes sense when the alternative is paying per-seat for a large support team plus the AI on top. At that scale, resolution-based pricing is the cheaper option.
Fin also benefits from Intercom's maturity around human-agent workflows. If your use case is "AI deflects easy tickets, humans take the hard ones" and the human workflow is your real product, Intercom's tooling is well built for that handoff.
Where Fin stops making sense is when the AI itself is the product you're buying — not a deflection layer on top of a shared inbox. If your team is 1–5 people, you don't already run Intercom, and what you want is a customer-facing AI that talks to leads and books meetings, the Fin stack is overkill and overpriced for the job.
If you're evaluating Fin seriously, write the answers to these into the contract:
None of these are gotcha questions. They're standard vendor diligence. If the answers aren't clean, the pricing model probably isn't right for your volume.
Per-resolution sounds like the fairest AI billing model until you multiply the rate by your real volume. At 3,000 conversations a month, Intercom Fin lands around $2,970 — before Intercom seats, before voice. Kazozo's Pro plan at $999 covers the same workload with cheap, predictable overages, no seat minimums, and voice at a simple $1.10 per answered call.
If you want to see the numbers on your actual traffic, the ROI calculator plugs in your volume and shows monthly costs on Starter, Growth, and Pro. Or book a 20-minute demo and we'll walk through a side-by-side against whatever Intercom quoted you.