Most support vendors charge a per-agent seat fee and a per-resolution AI fee. When AI resolves a ticket, you pay the resolution charge on top of seat capacity you already bought for the agent who would otherwise have handled it. The two lines scale on different curves and never offset.
Key takeaways
- Seat fees price capacity. Resolution fees price outcomes. Paying both means paying twice.
- Automation reduces the work but not the seat bill, unless you reduce headcount.
- The two lines grow on different curves, which makes forecasting harder.
- Ask what the seat fee buys once AI handles most of the volume.
Look at the invoice structure of almost any support platform in 2026 and you find two lines: a per-agent seat fee and a per-resolution AI fee.
Zendesk charges Suite seats plus per-resolution AI. Intercom charges seats from $29 plus $0.99 per Fin resolution. Freshdesk charges per agent plus $29 per agent for Copilot plus metered AI Agent sessions. HubSpot charges seats plus $0.50 per resolved conversation. Salesforce charges licences plus conversations or Flex Credits.
Consider what those two lines are for.
What each line buys
A seat fee buys capacity. The right for one person to log in and handle work, whether they handle one ticket or four hundred.
A resolution fee buys an outcome. One customer issue solved.
Now consider a ticket resolved by AI. You pay the resolution fee for the outcome. You also already paid the seat fee for the agent who would otherwise have handled it, and that agent is still on the licence.
The work happened once. It was funded twice.
Why it does not self-correct
The obvious response is that you should reduce seats as automation takes volume. In practice three things get in the way.
Seats do not track volume, they track people. A team of eight handling forty percent less volume is still a team of eight until someone makes a headcount decision. The bill follows the org chart, not the queue.
Annual commitments. Seat licences are frequently annual. Automation deployed in March does not change a commitment signed in January.
Access rationing works against you. The sensible response to AI handling routine volume is to give more of the company visibility into the remaining customer problems. Seat pricing charges you for exactly that.
So the two lines grow on different curves. Seats track headcount and contracts. Usage tracks customer demand. Neither offsets the other, and forecasting requires modelling both.
The question to ask your vendor
Not “what is your AI rate”. Ask this instead:
As AI handles more of our volume, what is the seat fee buying?
It is a fair question and the answers vary. Sometimes the platform genuinely is the system of record, the workflow engine, the reporting layer, and the seat fee is buying real infrastructure. Sometimes the honest answer is that it is buying the historic pricing model.
The follow-up: do you offer reduced-cost seats for people who need visibility but not full agent capability? Several vendors have light or viewer roles. Using them is usually the largest single saving available inside a seat model, and a lot of teams never ask.
How to compare across models
The comparison that works is total cost divided by issues genuinely resolved, including every line.
Worked example with illustrative figures. Ten agents, three thousand monthly AI resolutions:
- Seats at $55 each: $550
- AI at $1.50 per resolution: $4,500
- Total: $5,050
Versus a model with no seat fee at $0.59 per resolution:
- Seats: $0
- AI: $1,770
- Total: $1,770
Those numbers are illustrative and the structure is the point: at ten agents the seat line is small, and it is the interaction of rate and volume that dominates. Run it at your own seat count and volume before drawing conclusions, because at thirty agents and six hundred resolutions the picture looks completely different.
We published every published rate with the seat fees stacked if you want the full table.
Our position
Fidiora has no seat fee. Add your whole company, including engineers and account managers who should see customer problems, at no cost. You pay $0.59 per genuine resolution and nothing else.
That is a deliberate structural position rather than a discount: if you are paying for outcomes, charging again for how many people can log in is funding the same work twice.
Whether it is cheaper for you is arithmetic, not rhetoric. Run the comparison at your numbers.
Frequently asked questions
Why do support vendors charge both seats and usage?
Does automation reduce my seat costs?
How do I compare seat-plus-usage against usage-only pricing?
Is there a support vendor with no seat fee?
Resolve, don't deflect.
See Fidiora resolve a ticket, capture a lead, and keep the bill predictable.