Per-Resolution vs Flat Pricing: Which AI Support Model Actually Works?
Per-resolution pricing ($0.99 to $2.00 per ticket the AI closes) sounds fair until your ticket volume moves. Flat pricing does not change. Here is when each actually wins, with real numbers at 500, 2,000, and BFCM-spike volumes, and the hidden cost nobody prices in.
The two pricing models, stripped down
There are really only two ways AI support vendors charge. Everything else is a variation on these two.
Per-resolution
Pay a fee every time the AI resolves a ticket.
Intercom Fin: $0.99/resolution. Zendesk AI: $1.50-$2.00/resolution depending on plan. Usually stacks on top of a per-seat or platform base fee.
Flat monthly
Fixed fee for a conversation allowance.
Jarvis: $19 for 100 conversations, $59 for 500, $199 for 2,000. Unused conversations roll over. Spikes covered by flat-rate booster packs, not per-ticket charges.
The real math at different volumes
Using Intercom Fin's public pricing ($0.99 per resolution + $29/seat base) against Jarvis's public plans. Both vendors publish these numbers, so this is an apples-to-apples comparison.
100 conversations / month
Early-stage, low support volume
500 conversations / month
Growth-stage, steady volume
2,000 conversations / month
Mid-market steady state
2,000 -> 10,000 conversations (BFCM spike)
Peak season, 5x normal volume
Zendesk AI with per-resolution + per-seat + AI add-on runs even higher - roughly $750+/month for equivalent volume. Math is in the pricing page comparison.
When per-resolution pricing actually fits
Per-resolution is not always wrong. For a narrow set of patterns it genuinely makes sense.
Extremely low volume (under 50 tickets/month)
At low volume, per-resolution can work out cheaper than any flat plan because you pay only when AI engages.
Pilot or proof-of-concept only
Testing the waters with no commitment. If you are pausing after 30 days, you may never hit flat-plan breakeven.
Extremely predictable, flat-line volume
If your ticket count is identical month to month with zero growth and zero seasonality, you can model the cost precisely.
When flat monthly pricing fits
If any of these patterns describe your business, flat is the stronger model. Most SaaS, ecommerce, and mid-market teams fall into at least two of them.
Growing volume (adding customers or users)
Every new customer adds tickets. Per-resolution means your AI bill grows in lockstep with your customer base. Flat pricing decouples them.
Seasonal or spiky (ecommerce, SaaS with events, edu)
BFCM, launch weeks, product recalls, outage weeks. Per-resolution punishes the months you most need support to work. Flat absorbs them.
You want to invest in a great knowledge base
Per-resolution pricing creates a perverse incentive: the better your AI works, the more resolutions it handles, the higher your bill. Flat pricing rewards you for better KB inputs.
You need a budget CFOs can sign off on
"$2,000 this month, could be $8,000 next month" is a hard sell. "$199/month" is not.
The hidden cost of per-resolution pricing
The headline rate is $0.99. The actual cost shows up in the behavior changes the pricing creates inside your team.
Rationing behavior
Teams on per-resolution plans start gating what the AI can touch. Disable the AI on edge cases to save money. The AI becomes less useful over time because the pricing fights against usage.
Budget anxiety on great months
A viral moment or traffic spike is supposed to be a win. With per-resolution pricing, your first reaction is a pit in your stomach about the bill. That is a broken incentive.
Slower knowledge-base investment
A better KB means more AI resolutions. More resolutions means a higher bill. Teams on per-resolution pricing unconsciously slow down KB work because the feedback loop punishes them.
Forecasting overhead
Finance has to re-forecast support spend every quarter based on projected ticket volume. Flat pricing turns a line-item projection into a fixed subscription. One less thing to model.
How to audit which model fits you
You do not need a pricing consultant. A 15-minute audit on your own data tells you which model is cheaper and which one your team will actually thrive on.
01
Pull the last 12 months of ticket volume
Month-by-month total. Look at the highest month, the lowest month, and the average. The wider that spread, the more flat pricing wins.
02
Estimate AI resolution rate
What share of your tickets would an AI agent realistically close? A reasonable starting assumption is 60-70% for most SaaS and ecommerce, higher for very repetitive queues.
03
Run the math on both models
Multiply resolution count by $0.99 (per-resolution) and compare against the flat plan tier that covers your monthly volume. Do it for your peak month, not your average.
04
Add the hidden costs
How would your team behave if every AI resolution cost money? Would you ration it? Would your KB investment slow? Those have real dollar impact over 12-24 months.
The bottom line
Per-resolution pricing is a vendor-friendly model dressed up as customer-friendly. It benefits the vendor on every growth curve, every seasonal spike, and every KB improvement you make. Flat pricing stays out of the way. For any business that is growing, seasonal, or serious about knowledge-base investment, flat is the model that aligns incentives and keeps support a fixed line on the budget.
Common questions
What is per-resolution pricing for AI support?
A model where the vendor charges a fee for each ticket their AI successfully resolves. Intercom Fin charges $0.99 per resolution. Zendesk charges $1.50-$2.00 per resolution depending on plan tier. A resolution is usually defined as the AI answering without a human taking over.
What is flat monthly pricing for AI support?
A fixed monthly fee that includes a conversation allowance (e.g. $59/month for 500 conversations). The bill does not change based on how many tickets the AI resolves within that allowance. Overages are handled via top-up packs, not automatic per-resolution charges.
Which model is cheaper for most teams?
Flat is cheaper for almost any team with more than 100 tickets a month. At 500 conversations, per-resolution pricing costs roughly 9x flat. At 2,000 conversations, it is about 10x. Per-resolution only wins at very low volumes (under ~50 tickets/month) or in pure pilot mode.
Do per-resolution vendors offer discounts at volume?
Sometimes, through negotiated enterprise contracts. But those typically have annual commits and still scale with volume. Public list prices are what most teams actually pay in the first year.
What happens during a traffic spike or BFCM?
On per-resolution pricing, a 5x spike means a roughly 5x bill that month. On flat pricing, you keep your flat rate; overflow is handled via booster conversation packs (typically $9 per 50 conversations, far cheaper than $0.99 x 50). Flat pricing is the stronger fit for any business with seasonal or event-driven volume.
Does per-resolution pricing align incentives with the customer?
Vendors argue yes: they only get paid when the AI works. In practice, the vendor makes more money when you have more tickets, which creates a soft disincentive against helping you reduce ticket volume at the source (better KB, better product docs, better UX). Flat pricing removes that conflict.
Flat monthly pricing. No per-resolution surprises.
$19 for 100 conversations. $59 for 500. $199 for 2,000. Same bill, every month. Free plan: 10 conversations, forever.
No credit card required. Setup takes under 10 minutes.
Questions? Reach out at hello@deskclone.ai