Per-resolution vs flat pricing: break-even
Published by Elite Support AI, so not neutral. Every competitor figure below links to its source and the date it was read; figures that could not be confirmed on that date are left out.
Two pricing models dominate AI customer support. Per-resolution pricing charges each time the AI resolves a conversation. Flat pricing charges a monthly fee for an allowance. Neither is better in general; the crossover depends on how many conversations you handle, and it is lower than most buyers expect. This page works the arithmetic with published numbers.
The two models in one line each
| Model | You pay for | Example | Source |
|---|---|---|---|
| Per resolution | Each conversation the AI resolves | Intercom Fin: $0.99 per outcome (separate from Intercom seat plans) | intercom.com/pricing, read 2026-10-28 |
| Flat | A monthly allowance | Elite Support AI: $29, $79 or $199 a month for 1,000, 4,000 or 15,000 AI replies | Pricing section, 2026-10-28 |
| Per seat | Each human agent | Zendesk Suite Team: $55 per agent per month, billed annually (AI agents are priced separately on the page) | zendesk.com/pricing, read 2026-10-28 |
Break-even, worked
At $0.99 per resolution, the monthly cost of resolutions alone is 0.99 times the count. The break-even against each flat plan is the plan price divided by 0.99:
| Flat plan | Price | Break-even resolutions per month at $0.99 |
|---|---|---|
| Starter | $29 | 30 (29 / 0.99 = 29.3) |
| Growth | $79 | 80 (79 / 0.99 = 79.8) |
| Scale | $199 | 202 (199 / 0.99 = 201.0, rounded up) |
Below 30 resolutions a month the per-resolution model is cheaper than Starter. Above 80 it costs more than Growth, if Growth's allowance covers your volume. The comparison is on the AI line only: Intercom charges seats separately, and a helpdesk is a different product from an AI add-on.
Replies are not resolutions
Flat plans count AI replies, and one resolved conversation often takes several. If you assume three replies per resolution, the allowances translate into roughly these conversation counts:
| Plan | AI replies | About this many conversations at 3 replies each |
|---|---|---|
| Free | 50 | 16 |
| Starter | 1,000 | 333 |
| Growth | 4,000 | 1,333 |
| Scale | 15,000 | 5,000 |
At 500 resolutions a month, per-resolution pricing at $0.99 is about $495. Under the three-reply assumption, 500 conversations is 1,500 replies, which sits inside Growth at $79. At 2,000 resolutions the per-resolution figure is about $1,980 and the flat equivalent is Scale at $199, provided 6,000 replies are right for your conversations.
What favours per-resolution pricing
- Very low or unpredictable volume.
- A rule that you pay only when the AI actually resolves something.
- Needing the vendor's wider helpdesk anyway.
What favours flat pricing
- A predictable budget with no month-end surprise.
- Volume above a few dozen conversations a month.
- Seasonal spikes, where per-resolution bills rise with demand.
Risks on each side
Per-resolution models depend on what counts as a resolution, and definitions are set by the vendor; read them. Flat plans cap what the AI answers: when the allowance runs out, a person answers instead. That is predictable, but it means you must staff for overflow. Use the cost calculator to plug in your own numbers.
A decision rule you can apply today
Count last month's conversations that a machine could plausibly have answered, multiply by three replies each, and compare the total with the plan allowances above. If the number is under 50, the free plan covers it. If a per-outcome price times your conversation count is under $29, per-outcome billing is cheaper on the AI line alone. Otherwise a flat plan is cheaper, and the gap widens quickly with volume: at 1,000 conversations, per-outcome billing at $0.99 is $990, while the flat plan that holds 3,000 replies is $79. Recheck the numbers whenever either vendor changes its page, because both models can be repriced.
One more cost: the humans
Neither model removes people. Someone must review edge cases, update the source content and handle escalations. Budget a few hours a week at the start. Counting that time makes the flat plan look even better when volume is high, and makes low-volume per-outcome billing look better when you only need the AI for a handful of questions.
Frequently asked questions
What is a resolution?
The vendor's definition of a conversation the AI completed without a human. Definitions differ, so read the vendor's own page.
Does the flat plan charge for overages?
No. Past the allowance a person answers rather than the AI, and nothing is billed per message.