Per-agent vs per-resolution pricing: modelling 12 months of support cost
Two billing shapes, three worked scenarios and the exact ratio where each one starts to win. Real arithmetic on published rates, including what a peak month does to a variable bill.
Support software used to be priced one way: per seat, every month, whether the queue was busy or empty. Now most vendors bill the AI separately and charge for volume. The headline rate is the part everyone compares and the part that tells you least. What decides the invoice is which events trip the counter, how much allowance comes with each seat, and what your volume does in November.
The two shapes, and why the rate is not the model
A fixed cost is easy to budget and indifferent to how well you do. If the AI resolves most of your queue this month, a per-seat bill does not move; you keep the saving as freed-up agent time rather than as money back. A variable cost does the opposite: it falls in a quiet quarter and rises exactly when you are busiest, which is when a surprise costs the most.
Most vendors now sell a hybrid of the two, with the components hidden inside one monthly figure. These are the shapes you will be quoted, in the units each vendor publishes.
The billing shapes behind the headline rates
Shape
What you pay for
A published example
Seats only
Configured users, regardless of volume
The classic helpdesk model: fully predictable, and it does not fall when automation removes work.
Seats plus an AI allowance and packs
A seat price, an included AI quota, then blocks of extra capacity
LiveChat's Growth tier at $79 per agent per month billed annually includes 200 AI-agent resolutions, then packs of 50 at $49.50. Freshdesk's Growth tier at $19 per agent per month annually includes the first 500 Freddy sessions, then 100-session packs at $49.
Volume bundles plus a per-conversation surcharge
A conversation bundle for the whole workspace, then a fee per AI-handled conversation
Trengo's Boost at €349 per month covers 10 users and 500 conversations, with extra conversations at €18 per 100 and an AI surcharge of €0.30 per AI-handled conversation.
Volume only, no seat component
Tickets, with the plan sized by ticket allowance
Gorgias sizes plans by billable tickets — $60 for 300, $360 for 2,000, $900 for 5,000 — and adds $0.90 per resolved conversation on those tiers.
Seats plus a per-agent resolution allowance
A seat price whose included AI allowance also scales per agent, then a per-resolution rate
CustomerEagle: €24.99, €49.99 or €84.99 per agent per month, with 10, 25 or 50 AI resolutions included per agent per month, then €0.99, €0.56 or €0.49 per resolution above that.
Third-party figures were read from each vendor's own public pricing page on 26 and 27 July 2026 and are quoted in the currency that page publishes; they are not converted, and figures in different currencies are not compared against each other. Vendors change prices without notice — check the source before relying on any figure. CustomerEagle is not affiliated with, endorsed by, or partnered with any vendor named here; all product names are the trademarks of their respective owners.
Read that as different counters rather than different prices. A per-session counter trips more often than a per-resolution counter, so a lower session rate can produce a higher invoice at identical performance. That is a definition problem, not a pricing problem, and it is unpacked vendor by vendor in what counts as an AI resolution.
Five inputs, and one trap
Seats in twelve months, not today — including the part-time and seasonal ones, because most tools bill a seat the same either way.
Conversations in a real month. Not an average of a good one and a bad one.
The share answerable from documented knowledge. People guess at this and it moves the answer more than anything else. Count it on fifty real tickets.
Growth. A per-seat model reprices when you hire; a per-resolution model reprices every month.
Your peak multiplier, from your own history. In e-commerce it is usually the largest term in the model.
Scenario one: a three-agent webshop
Three agents, roughly 1,000 conversations in a normal month, of which the AI closes 550 without a human. November runs at 2.2 times normal, December at twice, January carries the returns wave at 1.3 times. Professional costs least at that ratio: €49.99 per agent per month, 25 resolutions included per agent, €0.56 after that.
Twelve months on Professional, three agents
Month
Resolutions
Seats (3 × €49.99)
Above the allowance (75)
Month total
Each of nine normal months
550
€149.97
475 × €0.56 = €266.00
€415.97
November (2.2×)
1,210
€149.97
1,135 × €0.56 = €635.60
€785.57
December (2.0×)
1,100
€149.97
1,025 × €0.56 = €574.00
€723.97
January returns wave (1.3×)
715
€149.97
640 × €0.56 = €358.40
€508.37
Twelve-month total
7,975
€1,799.64
€3,962.00
€5,761.64
CustomerEagle figures computed from the per-agent prices, per-agent allowances and per-resolution rates published on the pricing page; volumes are illustrative and should be replaced with your own. Prices exclude VAT.
Three things fall out of that. Seats are €1,799.64 of a €5,761.64 year, under a third of it. The three peak-affected months carry €2,017.91 — over a third of the annual bill in a quarter of the calendar. And Standard, whose seat price is half as much, would have cost €8,438.49 across the same twelve months, because its €0.99 rate applies to nearly every resolution once the 10-per-agent allowance is gone. A seat price €25 lower per agent bought a bill €2,676.85 higher.
Premium is worth checking too: it is lower than Professional in November and December — €774.37 against €785.57, €720.47 against €723.97 — because 50 included resolutions per agent and a €0.49 rate bite harder at peak. Switching plan for exactly those two months saves €14.70 over the year. That is the honest size of the plan-optimisation prize at this scale. Pick the plan that fits your normal month and leave it alone.
Scenario two: a ten-agent scale-up
Ten agents, no seasonal peak, volume compounding at 5% a month from 2,400 AI-closed conversations in month one. This is the shape where a variable bill hurts, because the thing driving it grows on its own.
Ten agents, volume growing 5% a month
Month
Resolutions
Lower-cost plan
Seats
Above the allowance
Month total
1
2,400
Professional
€499.90
2,150 × €0.56 = €1,204.00
€1,703.90
6
3,063
Professional
€499.90
2,813 × €0.56 = €1,575.28
€2,075.18
9
3,546
Premium
€849.90
3,046 × €0.49 = €1,492.54
€2,342.44
12
4,105
Premium
€849.90
3,605 × €0.49 = €1,766.45
€2,616.35
Twelve-month total
38,200
Switch in month 9
—
—
€25,620.99
CustomerEagle figures computed from the per-agent prices, per-agent allowances and per-resolution rates published on the pricing page; volumes are illustrative. Prices exclude VAT.
Staying on Professional all twelve months costs €25,710.80, so the mid-year switch is worth €89.81 — again, small. What is not small is the shape of the year: the monthly bill rises 54% between month one and month twelve while headcount does not move at all. On a pure per-seat contract that growth would have cost nothing extra, and the AI would have earned you nothing back. Both statements are uncomfortable, and choosing between them is the actual decision.
The crossover, stated as a ratio
Because the allowance on our plans scales per agent, the point where one plan overtakes the next does not depend on team size. It is a ratio — resolutions per agent per month — and it comes out the same for a solo founder and a fifty-person team.
Where each step up starts to pay for itself
Step
Monthly billing
Annual billing
Standard → Professional
About 49 resolutions per agent per month
About 37
Professional → Premium
350 resolutions per agent per month
About 279
Derived from the per-agent prices, per-agent allowances and per-resolution rates published on the pricing page. Annual billing lowers the seat price and leaves the per-resolution rate unchanged, which moves both crossovers down. Prices exclude VAT.
A one-agent workspace closing 60 resolutions a month is already past the first crossover: €74.49 on Standard against €69.59 on Professional. That is the counter-intuitive part — the plan with the higher seat price can be the lower total from a very small volume onwards, because the seat comes with a cheaper unit rate. The same arithmetic works on any vendor that publishes a computable formula, which our comparison pages do; where a vendor documents a mechanic without a rate, no figure gets invented for it.
The variables outside the rate card
Does the included allowance scale with seats? Ours does, so a five-person team on Professional starts each month with 125 resolutions before anything is charged. A flat workspace allowance behaves very differently as you hire.
Is there a monthly minimum? Some standalone AI products carry one, which turns a variable cost back into a fixed one at low volume.
Are the AI and the helpdesk billed separately, and can one conversation generate a charge in both?
What is the unit — a session, a conversation, or a resolved conversation? A session counter trips on every attempt, including the failures.
What does the free tier let you run? Ours is a testing tier, honestly: €0, one agent, and a one-time grant of 25 AI resolutions rather than a monthly allowance — enough to check your content against real questions, not enough to run a queue on.
Which downside would you rather own?
Per-resolution punishes a volume spike that arrives without matching content: the counter runs on every conversation the AI handles, including the ones it handled badly, unless the definition excludes them. Per-seat punishes a quiet quarter and, worse, punishes success — the better your automation gets, the more you pay per human conversation handled. For most teams a variable cost tied to a strict outcome is the easier risk to own, because it moves with revenue: a peak that doubles the support bill is a peak that doubled the orders. The volume side of that same week is in the peak-season preparation guide.
That only holds if the definition is strict, which is why two vendors quoting €0.50 per resolution are not quoting the same price when one of them never reverses a count. If one counted resolution in six turns out to be a conversation the customer reopened, a rate that never reverses is effectively 20% higher at the same headline number. Ours is written out on how we measure AI resolutions, and every counted resolution is visible with its full conversation history, so the invoice can be audited against transcripts rather than trusted. The plan rates used throughout are on the pricing page.
Is per-resolution pricing cheaper than per-agent pricing?
Not inherently. Per-resolution pricing converts a fixed headcount cost into a variable volume cost, so it favours teams whose ticket mix is repetitive and whose knowledge base is well documented, and penalises teams whose volume spikes without matching content. The deciding figure is resolutions per agent per month, which only your own data can tell you.
How do I calculate the cost of AI customer support over a year?
Use five inputs: seats in twelve months, conversations in a normal month, the share answerable from documented knowledge, your growth rate and your peak multiplier. Multiply conversations by the answerable share to get resolutions, apply each vendor's own formula month by month including the peak months, then compare twelve-month totals rather than headline rates.
How many AI resolutions are included per agent on CustomerEagle plans?
The allowance scales with seats: 10 AI resolutions per agent per month on Standard, 25 on Professional and 50 on Premium. Resolutions above the allowance cost €0.99, €0.56 and €0.49 respectively. The Free plan has no monthly allowance at all — it carries a one-time grant of 25 resolutions at signup instead.
At what volume should I move to a higher support plan?
On CustomerEagle plans the crossover is a ratio rather than a team size, because the included allowance scales per agent. With monthly billing, Professional starts costing less than Standard at roughly 49 resolutions per agent per month, and Premium overtakes Professional at 350. Annual billing moves both crossovers down, to roughly 37 and 279.
What happens to a per-resolution bill during peak season?
It rises in proportion to the volume the AI handles, which is also when revenue rises, but it can be a large share of the annual total. In the worked twelve-month model of a three-agent webshop above, the three peak-affected months accounted for just over a third of the year's software cost while representing a quarter of the calendar.
Does annual billing change per-resolution costs?
On CustomerEagle it discounts the seat price and leaves the per-resolution rate unchanged, so only the fixed half of the bill moves. That matters most for teams with low resolution volume per agent, and it also shifts the point at which stepping up a plan pays for itself, because the seat premium between plans shrinks while the rate gap stays the same.
Intercom, Zendesk, Gorgias, Freshdesk and Tidio all bill for AI outcomes — and all define them differently. A sourced, side-by-side look at what each one actually charges you for.
Deflection counts tickets that never arrived. Resolution counts problems that got solved. They are not the same metric, and only one of them survives contact with an angry customer.
Resolve more tickets automatically.
See how honestly-measured AI resolutions cut your support load — start on the Free plan, no credit card, no sales call to get started.