Here is the short version. Per-agent and per-fulfiller both mean you pay for the people who work tickets, not the end users who raise them, and the label matters far less than the usage definition the vendor attaches to it. Flat or unlimited models trade per-seat math for a predictable fee, which helps some orgs and overcharges others. The right model is the one that ties your cost to real work at your size, and the definition behind it is where the negotiation lives. This sits inside our 2026 ITSM platform cost comparison.
What is the difference between per-agent and per-fulfiller?
Functionally, very little. ServiceNow says fulfiller, others say agent, but both charge for staff who resolve tickets while leaving requesters free or low-cost. The real difference is in the definition each vendor bolts on: who counts as an agent, whether occasional or part-time staff are chargeable, and how service accounts are treated. That definition decides your bill more than the model name, which is why we treat it as the primary lever in usage-definition clauses.
When does a flat or unlimited model win?
Flat or tiered-band pricing wins in two cases. First, a small org with a stable, small team of agents, where a single predictable fee beats counting seats and the administrative drag of true-ups. Second, an org with a volatile agent count that would otherwise face painful true-ups every period, where a flat band buys predictability. It loses when your agent count is small relative to the flat fee, because you are then subsidizing capacity you do not use, which is the most common flat-model trap.
Which model wins by org size?
| Org profile | Usually wins | Why | Buyer-side lever |
|---|---|---|---|
| Small, stable team | Flat / tiered band | Predictable fee beats per-seat overhead | Confirm the band matches actual agents, not aspiration |
| Small, volatile team | Flat band with growth room | Avoids repeated true-ups | Cap the uplift and negotiate downward flex |
| Mid-market, growing | Per-agent / per-fulfiller | Cost tracks real working staff | Tight definition; reclassify requesters; ramp protection |
| Large enterprise | Per-fulfiller with right-sized counts | Ties cost to ticket-working staff; shelfware is reclaimable | Narrow definition, shelfware reclamation, AI scoped separately |
These are tendencies, not rules; the answer depends on your real counts, growth and the definition on offer, so model it rather than assume. As of 2026, AI is increasingly metered on top of whichever seat model you choose, so price it as its own line and do not let it ride inside the model comparison.
How do you negotiate the model itself?
- Reconcile your true agent count first. The model only pays off against an honest count, not the provisioned one.
- Negotiate the definition, not just the rate. Whoever counts as an agent decides the bill; pin it down narrowly.
- Test all three models on your numbers. Ask the vendor to quote per-agent and flat, then run your own math on both.
- Watch the flat-model floor. Unlimited only wins if you fill it; otherwise it is a subsidy you pay the vendor.
- Scope AI and add-ons separately. Keep metered AI out of the model comparison so it cannot hide the real seat cost.
Across ITSM engagements our firm averages a 30% reduction, with more than $420M negotiated since 2019 on a Map, Benchmark, Leverage, Close method, and choosing the right model for your size is one of the cleaner wins because it compounds across the whole term.
Our gated ITSM Competitive Leverage Playbook includes the model-comparison worksheet and the definition checklist behind this analysis.
Book a leverage review.
We model per-agent, per-fulfiller and flat on your real counts and pick the one that wins at your size, then negotiate the definition behind it. Fixed fee or gainshare.
Book a competitive leverage review →What to do next
Reconcile your real agent count, then ask your vendor to quote per-agent and flat side by side and run the math yourself before you accept the model as given. Pair it with a tight usage definition and the right-sizing covered in ServiceNow vs BMC Helix vs Ivanti. The engagement runs through our competitive leverage service on fixed fee or gainshare, and the full field is in the platform cost comparison pillar.
Frequently asked questions
- What is the difference between per-agent and per-fulfiller pricing?
- Usually the same idea under different names: you pay for staff who work tickets, not requesters. The usage definition each vendor attaches matters more than the model label.
- When does a flat or unlimited model win?
- For small stable teams where a predictable fee beats per-seat math, and for volatile counts that would otherwise true-up painfully. It loses when your count is small relative to the flat fee.
- Which model is cheapest for a large enterprise?
- Per-fulfiller or per-agent with a tight definition and right-sized counts usually wins, because cost tracks real working staff and shelfware is reclaimable. Model your own counts before assuming.