A support contract can hit every number it signed and still fail the business it serves.
The classic SLA measures the machinery of support: how fast a ticket was acknowledged, how fast it was resolved, how many were closed. All of it is worth knowing. None of it asks the question the business is actually asking — did the systems hold up when it mattered?
Response time measures the stopwatch. The business measures whether payroll ran, the month closed and the recurring failure finally stopped recurring.
Where the classic SLA goes blind
Averages hide the incident that mattered
A month of green averages can contain the one P2 that stalled payroll approval for a day. Statistically it drowned in forty routine tickets; commercially it was the month. Averaged metrics are designed — unintentionally — to dilute exactly the events the business remembers.
All hours are priced the same
An interface failure on an ordinary Tuesday and the same failure during the payroll run differ by an order of magnitude in consequence — and not at all in the classic priority matrix. Severity schemes that ignore the business calendar misprice the risk the contract exists to manage.
Closure is rewarded; prevention is invisible
Nothing in a response-and-resolution regime loses points when the same incident returns monthly — and nothing gains points for the root-cause fix that makes it never return. The contract literally cannot see the difference between a treadmill and an improvement curve.
The same month, two readings
What to put in the contract instead
- Business-moment availability — named critical windows (payroll runs, close, peak hiring) with their own protection, staffing and reporting, agreed on the business calendar.
- Impact-weighted severity — priority set by what the incident blocks and when, not only by how many users typed a ticket.
- A recurrence-retirement measure — the count of incident classes that existed last quarter and no longer exist, with capacity ring-fenced to drive it.
- Improvement capacity in the base fee — a protected slice of hours for root-cause work, so prevention is not the first casualty of a busy month.
Map the business moments→Weight incidents by them→Ring-fence improvement capacity→Report what stopped recurring
Keep the classic numbers — demoted
Response and resolution times still belong in the contract; they police the machinery. The change is hierarchy: they become hygiene measures beneath the outcome measures, not the headline above them. A provider uncomfortable with that inversion is telling you which of the two they intend to manage.
VISCAP perspective
An SLA is a statement about what the provider is paid to protect. Make sure it names the things your business cannot afford to lose.
We structure support engagements around the client’s calendar first — the payroll runs, the closes, the seasonal peaks — because that is where support earns or loses its keep. The ticket-queue numbers stay on the dashboard; they have simply stopped being the point.
One question for your current contract
Find last quarter’s report and ask: could these numbers all be green in a month where payroll went out late? If yes, the contract measures the queue — and the queue is not the business.