Flat vs Per-Resolution AI Support Pricing: Which Saves Money?

AI support tools now split into two pricing camps, and the one you pick quietly decides your bill for the next two years. On one side: per-resolution (or per-seat metered) pricing, where you pay each time the AI closes a ticket. On the other: flat monthly pricing, where one price covers a fixed conversation allowance no matter how many the AI resolves. Both are defensible. But they behave very differently as your volume grows, and the crossover point catches most teams off guard.

TL;DR

• Per-resolution pricing charges per ticket the AI resolves - Intercom Fin is the headline example, listed at *around* $0.99 per resolution plus a per-seat fee (roughly $29–$132/seat as publicly listed; subject to change).
• Flat pricing charges one predictable monthly fee for a conversation allowance. Chatonbo is flat: Free $0 (100 conv/mo), Pro $19/mo (5,000 conv), Business $59/mo (10,000 conv) - no per-resolution or per-conversation fees.
• Metered pricing can win at low, spiky volume - if you resolve a few hundred tickets a month, paying per resolution may cost less than a fixed plan.
• Flat pricing wins as you scale. Per-resolution cost rises linearly with traffic; flat cost stays put. The lines cross fast.
• At ~10,000 conversations/month, a per-resolution tool can land in the thousands per month (illustrative), versus Chatonbo's flat $59. We show the math below - treat the competitor figure as an estimate, not a quote.

The two AI support pricing models, defined

Per-resolution pricing charges a fee every time the AI fully answers a customer without a human. You also typically pay a per-seat license for the humans who handle the rest. Your bill = (resolutions x per-resolution rate) + (seats x seat price). It scales with usage.

Flat monthly pricing charges one fixed price for a defined allowance of conversations. Your bill = the plan price, full stop, until you exceed the allowance and move up a tier. It scales with *tiers*, not with each ticket.

The distinction matters because support volume is rarely flat. It spikes during launches, sales, outages, and holidays - exactly when a per-unit meter runs hottest.

What "per-resolution" actually costs (Intercom Fin as the reference)

Intercom Fin is the most-cited per-resolution product, so it's the useful reference point. As publicly listed, Fin runs around $0.99 per resolution, layered on top of a per-seat plan (seats listed in the $29–$132 range depending on tier). These numbers are illustrative, publicly posted, and change over time - always check the current pricing page before budgeting.

The appeal is real: you pay for outcomes. If the AI resolves nothing, you pay (almost) nothing beyond seats. For a small team with a few hundred tickets a month, that can be cheaper than committing to a fixed plan.

The trap is also real. "Resolution" is defined by the vendor, billing rises every time traffic rises, and a single viral moment or seasonal surge multiplies your bill with no ceiling. You're effectively renting your support capacity by the unit.

What flat pricing costs (Chatonbo as the reference)

Chatonbo is deliberately flat. Every plan includes the AI - there is no separate "AI add-on," no per-resolution meter, and no per-conversation charge:

• Free - $0/mo, 100 conversations/mo
• Pro - $19/mo, 5,000 conversations/mo
• Business - $59/mo, 10,000 conversations/mo

Agentic features - order tracking, product recommendations, one-click [checkout links](/glossary/agentic-commerce) - are included with no extra per-use charge. Your bill is the same in a quiet month and a record month. That predictability is the entire point: you can forecast support cost a year out without modeling traffic.

Worked example: the cost at different volumes

Here's where the models diverge. The formula for a per-resolution tool is:

Monthly cost = (conversations x resolution rate x $0.99) + seat cost

We'll assume ~50% of conversations become billable resolutions (a common rough split - many chats are duplicates, spam, or escalate to a human) and one seat at ~$39 for illustration. These are assumptions, not quotes - your real numbers depend on your traffic mix and the vendor's current rates.

*Per-resolution figures are illustrative estimates using ~$0.99/resolution, a 50% resolution rate, and one ~$39 seat. They are not vendor quotes and will vary with your actual rates and traffic.*

The pattern is unmistakable. At low volume the gap is modest. By 10,000 conversations a month, the per-resolution estimate sits in the ~$5,000/month range while flat pricing holds at $59. Even if your resolution rate is half our assumption, you're still comparing four figures to two.

So who should pick each model?

Be honest about your situation - neither model is universally right.

Per-resolution can make sense if:

• Your volume is genuinely low (a few hundred tickets/month) and stable.
• You want to pay strictly for outcomes and your traffic rarely spikes.
• You're piloting AI and want near-zero commitment while you test resolution quality.

Flat pricing wins if:

• Your volume is mid-to-high, or seasonal and spiky.
• You need to forecast support cost without modeling every ticket.
• You're scaling - growth shouldn't linearly inflate your support bill.
• You want agentic actions (order lookups, recommendations) without per-use surcharges.

The deciding question isn't "which is cheaper today?" It's "what happens to this bill when traffic doubles?" Under flat pricing, nothing happens until you cross a tier. Under metered pricing, your cost doubles with it.

The hidden-cost trap of metered pricing

Three things make per-resolution bills bigger than teams expect:

1. Volume is the multiplier you don't control. Marketing campaigns, product launches, and outages all spike tickets. Each spike is a billing event.
2. "Resolution" is defined by the vendor. A reopened ticket, a follow-up question, or a loosely-counted auto-reply can each register as billable. Small definitional differences move the total a lot.
3. Success raises your bill. The better your AI gets, the more it resolves, the more you pay. You're penalized for the exact outcome you wanted.

Flat pricing inverts that: a smarter AI that resolves more costs you nothing extra. The incentive points the right way.

FAQ

What is per-resolution pricing for AI support?

Per-resolution pricing charges a fee each time an AI agent fully resolves a customer ticket without a human, usually on top of a per-seat license. Intercom Fin is the best-known example, listed at *around* $0.99 per resolution plus seat fees (illustrative, subject to change).

Is flat or usage-based AI pricing cheaper?

Usage-based (per-resolution) pricing can be cheaper at low, stable volume. Flat pricing becomes cheaper as conversation volume grows, because per-resolution cost rises with every ticket while flat cost stays fixed within a tier.

How much does Intercom Fin cost?

As publicly listed, Intercom Fin is *around* $0.99 per resolution plus a per-seat plan (roughly $29–$132/seat depending on tier). These figures are illustrative and change over time - confirm on Intercom's current pricing page. See our [Chatonbo vs Intercom comparison](/compare/intercom) for a side-by-side.

Does Chatonbo charge per resolution?

No. Chatonbo is flat: Free $0 (100 conv/mo), Pro $19/mo (5,000 conv), Business $59/mo (10,000 conv). There are no per-resolution or per-conversation fees, and AI is included on every plan. See full [pricing](/pricing).

At what volume does flat pricing win?

There's no single number, but flat pricing typically pulls ahead once you're resolving more than a few hundred tickets a month. By ~10,000 conversations/month the gap is large - thousands per month versus a flat $59 in our illustrative example.

The bottom line

Per-resolution pricing is a pay-as-you-go bet that your volume stays small and steady. Flat pricing is a fixed-cost commitment that pays off the moment you scale. For most growing businesses - especially anyone with seasonal spikes or a campaign calendar - flat wins, and it wins by a wide margin at volume.

Run your own numbers before you commit. Plug your real monthly conversation count into the table above, then [try the Chatonbo pricing calculator](/pricing) to see your flat cost at every tier. When you're ready, [start free](/platform/register) - no credit card, and no meter running in the background.