Response times that mean something
A four-hour response commitment is worthless without a definition of when the clock starts, what counts as a response, and what happens when it is missed.
Service level agreements in maintenance contracts are usually a table of priorities and hours. They look precise and they are frequently unenforceable, because the terms in them are not defined. For a separate product-side perspective on coordinating commitments across several stakeholders, read more here provides additional detail.
Define the start of the clock
When does the four hours begin? When the fault occurred, when it was noticed, when it was reported, when the report was acknowledged, or when the contractor's system logged it?
Each is defensible and they can differ by hours. Without a stated definition, the contractor will use the latest one and the client will assume the earliest, and neither is being dishonest. State it, and state which clock applies out of hours — a fault reported at six on Friday evening under a working-hours SLA does not have a four-hour response, and both parties should know that before it happens. For a wider operational and compliance reference, consult HVACR maintenance and contracting resources.
Attending within four hours and fixing within four hours are different promises with very different prices. Conflating them is the single most common defect in maintenance SLAs.
Define what a response is
Is a phone call a response? An email acknowledgement? Remote diagnosis? An engineer on site?
All four are legitimate depending on the fault, and the contract should say which counts for which priority. Otherwise an acknowledgement email at three hours and fifty minutes satisfies the SLA while nobody has looked at the failed plant.
Rectification needs a stop-the-clock rule
Contractors reasonably resist open-ended rectification commitments, because some faults require parts on long lead times that are outside their control.
The workable structure is a rectification target with defined suspension conditions — awaiting a part on order, awaiting client access, awaiting client authorisation for a quotation — each of which stops the clock and each of which must be recorded with a timestamp. Suspensions that can be applied without evidence become the mechanism by which every target is met.
Measure it from your own records
SLA performance reported by the contractor from the contractor's system is not verification. It is not necessarily wrong, and it is not independent.
Log the report time and the attendance time in your own system. Where the two sets of figures diverge, that is a conversation worth having early, because it is nearly always a definitional difference rather than a dispute — and it is far easier to resolve in month two than at contract renewal.
Make failure consequential, proportionately
An SLA with no consequence is a statement of intent. Financial penalties are the usual instrument and are worth keeping modest and mechanical rather than large and negotiable — small, automatic deductions applied consistently change behaviour more reliably than a substantial penalty that is always waived.
The more effective consequence in a long relationship is usually procedural: a defined escalation, a review meeting triggered by repeated failures, and a documented record that feeds the renewal decision.
Do not buy a response time you do not need
Response times are priced. A four-hour commitment across all assets costs considerably more than a four-hour commitment on the assets where four hours matters.
Use the criticality bands from the asset register to differentiate. Most buildings have a small number of assets that genuinely warrant an urgent response and a long tail that does not, and paying the urgent rate across the whole estate is a common and expensive default.