Service Level Agreement: The Commitments, the Maths and the Exclusions

A service level agreement states measurable standards a supplier commits to, and the remedy when those standards are missed. It is a commercial instrument rather than a technical one: its purpose is to allocate the consequences of failure, not to prevent failure.

It is also not the same thing as reliability. A supplier with a demanding agreement and poor engineering will pay credits and still cost you working days. A supplier with a modest agreement and excellent operations may never trouble you. The agreement matters most as a signal of what the supplier is prepared to stand behind and as a mechanism when things go wrong repeatedly.

The formula

Availability equals total minutes in the period minus downtime minutes, divided by total minutes, expressed as a percentage. The formula is trivial. Everything interesting is in the definitions: what counts as downtime, whether partial degradation counts, who measures, and from which vantage point.

Converting percentages into minutes

A thirty-day month contains 43,200 minutes and a year contains 525,600. Applying the target percentage to those totals converts an abstract figure into a number of minutes you can reason about.

The two columns explain why the measurement period is not a detail. A monthly 99.9 per cent commitment caps any single month at roughly forty-three minutes. The same percentage measured annually permits nearly nine hours, potentially all at once, without the commitment being missed.

Support commitments are normally tiered by severity, and the severity definitions carry most of the weight. A definition that reserves the highest tier for complete unavailability affecting all users will rarely be met by real incidents, most of which affect some users or some functions. Definitions that include substantial degradation are more useful and harder to obtain.

Within each tier, distinguish three separate commitments: acknowledgement, update frequency during an incident, and resolution. Most agreements commit firmly to the first two, which is reasonable, since resolution time depends on what has broken. Update frequency is the underrated one. During a serious incident, a commitment to a status update every thirty minutes is often worth more to a customer than an optimistic resolution target, because it lets them manage their own business around the outage.

Service credits are usually calculated as a percentage of fees for the affected period, on a scale that rises with the size of the shortfall, and capped at some portion of those fees. They are compensation rather than damages, and they are frequently stated as the sole remedy, which means accepting the credit closes the matter.

Exclusions are where the real commitment level is set. Scheduled maintenance, force majeure, customer-side failures and third-party network problems are all standard and defensible. What matters is whether they are bounded. A maintenance exclusion with no cap on hours and no notice requirement can absorb a large share of the period. A third-party carve-out drafted broadly can exclude much of what a customer would consider an outage.

Finally, read the claim process. Many agreements make credits available on request only, within a window measured in days after the affected period. A commitment that requires the customer to notice, calculate and apply within a fortnight will often produce nothing at all, not because it was missed but because nobody was assigned to claim it.

An uptime commitment is the availability component of a service level agreement considered on its own. Service credits are the remedy. A master service agreement usually contains the framework within which the service levels sit. An opt-out clause conditioned on repeated service failures is the escalation beyond credits, and it is the provision that gives an agreement genuine force. Nothing here is legal advice, and service level terms are worth qualified review on any contract of significance.

Related pages on evaluating and running a CRM:features overview, book a demo, pricing, what a CRM is, compare alternatives, and CRM for small business.