Field-Service Advisory

The SLA premium illusion: why faster response can destroy margin

A faster response promise is not just a better service level. It is an operating constraint — it removes routing freedom and forces you to hold capacity in reserve. Price that, or the premium quietly loses money.

By Paula Navarro · Cosmicalley  |  July 2026  |  8 min read

Field service is being squeezed from both sides. In Simon-Kucher's field-services research, more than 80% of firms name cost pressure and labour shortages as their biggest challenges — alongside intensifying competition and, tellingly, gaps in pricing discipline. The commercial reflex is to move up the value ladder: package tiered service-level agreements and charge more for faster response. Two-hour, four-hour, same-day, next-business-day — each a higher price point.

It is a reasonable instinct. But it rests on a quiet assumption: that a faster promise is a better version of the same service, sold at a markup. It isn't. A response window is not a feature you bolt on — it is a constraint you impose on every day you will ever schedule, and its cost lands somewhere the price rarely looks.

You didn't sell a faster van. You sold a constraint.

When you promise a customer a two-hour response, you have not added a resource. You have removed an option. From that moment, the scheduler can no longer treat that customer's work as something to be grouped with nearby jobs, deferred to a fuller run, or shifted to balance a technician's day. The job now carries its own clock, and the whole operation has to be arranged so that clock can always be honoured — including on the days the call never comes.

A response promise is priced as a service level. It behaves as an operating constraint.

Why a tighter window costs more than a faster one

The mechanism is routing freedom. Operations research has modelled this for decades: the vehicle-routing problem with time windows is, at heart, a study of what happens to cost when you narrow the interval in which each customer must be served. Narrower windows shrink the set of feasible routes, and the objective the literature optimises is an explicit trade-off between travel distance, waiting and idle time, and the number of vehicles required. Tighten the windows and you can drive up travel, waiting, idle capacity or fleet size — often several at once.

In an operation, that reads concretely. With broad windows, a dispatcher batches four jobs in one neighbourhood into a single efficient loop and lets a fifth wait for tomorrow's run past the same street. Impose two-hour promises on the same customers and the batching collapses — each job must be reachable inside its own window, so routes zigzag across the territory, vans travel further, and a van has to sit unassigned so that coverage exists if a call lands. The cost of response is not linear in speed. It rises as the window closes, because what you are really buying is the freedom you gave away.

Response is not resolution

There is a second trap, and it is a pricing trap disguised as an operational one. Response time and resolution time are distinct measures — IBM, whose service documentation defines the terms, treats them as separate SLA metrics: response is how quickly you acknowledge and arrive; resolution is how long until the asset is actually working again.

You can meet the response SLA and still fail the customer outcome. If the technician arrives without the right part, the right skill, or the right diagnostic information, the clock was beaten and the machine is still down. And the premium leaks money twice: you paid for the speed — the reserved capacity, the extra travel — and then, because the job wasn't ready to be resolved on arrival, it becomes a repeat visit, which consumes that capacity again. The customer bought restored operation. You delivered a fast arrival and a return trip.

Same jobs, different promise

Hold the work fixed and change only the promise, and the operating picture changes underneath it.

The most expensive part of the promise is the capacity you never use

Here is the line most SLA pricing misses. The cost of a premium response is not mainly the visit — the visit is marginal, and you would have made it under any tier. The expensive part is the capacity you must hold in reserve in case the incident occurs, on every day it doesn't. A tight SLA is not the sale of a service; it is the sale of an option — the customer's right to pull a technician inside two hours — and options have to be underwritten with standing capacity whether or not they are ever exercised.

Which means the only question that matters is an incremental one: does the premium cover the cost the promise adds? Put those two things side by side and the illusion becomes visible.

This is also where technology is often misdirected. Better routing optimisation does help — it finds the best grouping the constraints still allow, and raises the ceiling on how much work a tight day can hold. But no solver removes the reservation. A two-hour promise still needs slack held idle to be credible; optimisation makes the reserved capacity smaller, never zero. The lever that changes the economics most is not a smarter route. It is deciding which customers should carry a tight promise at all.

Not every asset deserves the same promise

Because the cost of a promise varies by where and what you're serving — and so does the value the customer gets from it — a single SLA tier applied across a book of contracts is almost always mispriced somewhere. Six dimensions decide whether a tight window earns its premium.

When premium response is absolutely right

None of this is an argument against fast SLAs — it is an argument for aiming them. Siemens' 2024 True Cost of Downtime study puts the combined cost of unplanned downtime across the world's largest firms in the trillions of dollars a year. And — more usefully for this decision — the per-hour cost of downtime varies by orders of magnitude across sectors: modest for a low-criticality asset, into the millions per hour on a semiconductor tool or an automotive line. That spread is the whole point. There are assets where the cost of downtime justifies a very substantial premium for reserved response capacity: a critical production line, a healthcare device, a safety-related system. There, reserve the capacity and price the promise to cover it. The mistake is not offering premium response. It is offering it uniformly.

Price the promise before you sell it

The discipline the reflex skips is simple to state and easy to get wrong by intuition: separate the cost of the visit from the cost of the reservation, work out what each SLA tier actually adds to cost-to-serve, and set the break-even premium before the tier goes on a rate card — then check how fast it erodes if incidents come more often than assumed. That is arithmetic, not judgement, and it is what the tool below does.