Per-Resolution vs Per-Seat AI ITSM Pricing: How to Compare

Compare per-resolution and per-seat AI pricing by normalizing both to annual cost at your own ticket volumes, then stress-testing each at plus 25 and plus 50 percent usage. Per-seat pricing (such as Atlassian Rovo at about $20 per user per month) is bounded by headcount and barely moves under that stress. Per-resolution pricing, where you pay for each ticket an autonomous agent deflects or resolves, is bounded only by volume and can climb steeply. The model you can defend is the one whose worst case you can afford.

What is the difference between per-resolution and per-seat AI pricing?

Per-seat pricing charges a flat monthly fee for every licensed user, whether they invoke the AI once or a thousand times. Per-resolution pricing charges per outcome, typically each ticket an autonomous agent closes without a human, so the bill tracks volume rather than population. The first is a fixed cost you size by headcount; the second is a variable cost you size by ticket flow. As of 2026 pricing, vendors mix the two, with copilot features sold per seat and autonomous resolution sold per outcome.

How do you compare the two on real numbers?

Run a four-step normalization rather than trusting either headline.

1. Fix your inputs. Pull your annual ticket volume, the share genuinely automatable, and the seat count that would carry a per-seat license.
2. Build both bills at expected volume. Per-seat equals seats times monthly rate times twelve. Per-resolution equals automatable tickets times the unit price.
3. Stress the per-resolution bill. Re-run it at plus 25 and plus 50 percent volume. The spread is the risk a cap has to cover.
4. Compare worst case, not expected. Choose on the number you can defend in a bad quarter, then negotiate the cap that makes the variable model behave like a fixed one.

What does the vendor argue, and what do you counter with?

The vendor selling per-resolution will frame it as paying only for value delivered, which sounds fair until volume scales and the meter has no ceiling. Counter by accepting the outcome framing and attaching a cap: a not-to-exceed annual figure, a fixed unit that cannot reprice mid-term, and an exclusion for reopened or disputed resolutions. The vendor selling per-seat will frame it as predictable, which is true, but push back on idle seats by negotiating the right to true down at renewal based on measured invocation, not just true up. The discipline mirrors the wider 2026 agentic AI ITSM pricing guide, and the cross-vendor view sits in Rovo vs Now Assist vs Freddy.

What to do next

Build both bills before you choose a vendor, not after. The normalization above takes an afternoon and changes which model you ask for, which is usually worth more than any discount you would win on the wrong one. When you want it run end to end, our AI cost control service models the crossover and negotiates the cap on fixed fee or gainshare.

Frequently asked questions

Is per-resolution AI cheaper than per-seat?

At low or seasonal automation volume, usually yes, because you pay only for outcomes. As volume scales it can overtake per-seat, which is bounded by headcount. Find the crossover volume and judge whether your automation rate will pass it within the term.

How do I cap a per-resolution AI contract?

Negotiate a not-to-exceed annual figure, a fixed unit price that cannot reprice mid-term, and an exclusion so reopened or disputed resolutions are not billable. That makes a variable meter behave like a budgeted line.

Which model is more predictable?

Per-seat, because cost is fixed by the number of licensed users. Per-resolution moves with ticket volume, so it needs a cap to become predictable.