Per-Seat vs Per-Transaction Pricing

Every outsourcing contract settles one question before any other: is the client buying people, or buying work? Per-seat pricing sells capacity. Per-transaction pricing sells volume. The choice reads like a commercial detail, yet it quietly decides what both sides measure, manage and ignore for the life of the deal.

Two models, two units of sale

Per-seat pricing (often FTE-based pricing) charges a fixed fee per agent or seat per month, from a rate card by role and location. The client buys capacity: a team of a certain size and shape, available for agreed hours. Invoicing is simple, revenue is predictable for the provider, cost is predictable for the client, and headcount is reviewed at agreed intervals. What the seats actually produce sits outside the commercial mechanism.

Per-transaction pricing charges for each unit of work processed: a call, a claim, an invoice, a case. The client pays for consumption; the provider carries the risk of idle capacity and earns from throughput. Because a pure pay-per-unit deal leaves the provider exposed to demand it cannot control, these contracts usually arrive with scaffolding: volume bands, minimum commitments, seasonal profiles and true-ups. The unit definition itself becomes a negotiated artefact, and often the most contested clause in the contract.

The incentive shape each creates

A pricing model is an incentive machine, and the two machines point in different directions.

Per-seat rewards inputs. The provider is paid whether the seats are productive or not, so the client carries utilisation risk: if volumes fall, the client pays for idle capacity. Worse, the model penalises improvement. A provider that automates a task or lifts productivity shrinks the number of billable seats, which means the better it gets at the work, the less it earns. Providers are rarely villains about this; the incentive simply removes the reward for efficiency, and organisations tend to do what they are paid to do. What per-seat makes invisible is output: the client can see exactly what a seat costs and almost nothing about what a seat produces, so productivity drifts unexamined and the cost of serving the process is anyone’s guess.

Per-transaction rewards throughput. The provider now profits from efficiency, which is the point of migrating. But the model makes different things invisible. Every unit earns the same fee, so the complexity mix disappears: a hard case and a trivial one are commercially identical, which invites cherry-picking the easy ones, fragmenting work into more countable units, and letting the difficult residue age. Quality becomes the pressure point, because faster processing earns more, and the service level agreement has to carry all the weight the price no longer does. Effort, judgement and the true cost of the hard cases all vanish behind the unit count.

A pricing model is not a commercial detail. It is a standing decision about who carries which risk, and it keeps deciding long after everyone has stopped looking at it.

One concrete example

Clearly illustrative, with no customer implied. An insurer outsources claims handling on a per-seat model. Each year the backlog justifies a few more seats, and each year the request is granted, because seats are the only unit anyone prices. Nobody measures claims per seat. When a new operations director finally derives a cost per claim, it has been rising for years while the invoice stayed respectable. The contract is migrated to per-transaction to fix this. Within two quarters the easy claims are flowing beautifully, the unit price looks excellent, and the complex claims are quietly ageing in a queue, because every one of them costs the provider more than it earns. The volume targets are met. The angriest customers are the ones whose claims were hard. Switching models did not remove the blind spot; it moved it.

The migration question

Most contracts that migrate move from seats to transactions as the work matures, and the migration is best understood as a repricing of risk, not a discount negotiation. The questions that decide whether it is safe: Is the unit of work definable, stable and countable by both sides? Is baseline productivity measured, or merely negotiated? Who keeps the gains when productivity improves, and what funds the automation that produces them? What happens to the complexity mix, and does the price recognise it, through tiered units or carve-outs? Are volumes forecastable enough for the provider to price the demand risk it is about to absorb? Where the answers are weak, hybrid structures (a capacity floor plus a variable unit fee, or unit pricing for the stable work only) let the contract migrate in stages instead of in one leap of faith. The same questions, asked honestly, sometimes point the other way: work that has become volatile or judgement-heavy can deserve to move back to capacity pricing, a pattern explored further in BPO vs BTO.

The decision-intelligence angle

The pricing model in a BPO contract is a decision made once, on assumptions about volumes, productivity and mix that begin decaying the day the ink dries. Most organisations then stop treating it as a decision at all: renewals roll the model forward because it is there, and the assumptions underneath are never re-examined. A decision-intelligence discipline treats the model as what it is: a standing allocation of risk that deserves to be revisited on evidence. That means recording what the model assumed when it was chosen, watching the assumptions that would change the answer (volume stability, complexity mix, measured productivity), and treating each renewal as a genuine decision with priced options rather than an administrative event. The firms that get hurt by pricing models are rarely the ones that chose badly. They are the ones that chose once.