Ridge ITIL

SLAs, OLAs, and the Art of Promising Well

A service level agreement is a promise with a measurement attached. Here's how promising works — and the watermelon trap that catches everyone.

Every mature service relationship eventually needs a written answer to an awkward question: exactly how good will this service be? "We'll do our best" satisfies nobody — the business can't plan around it and IT can't be fairly judged by it. Service level management is the practice of making promises properly: explicit, measurable, and — the hard part — actually keepable.

The three-letter family

Here's the discipline the family enforces, and where beginners' eyes should narrow: an SLA is only as strong as what underpins it. Promise customers 4-hour resolution while your hardware vendor's contract says next-business-day parts, and you've signed a promise arithmetic already broke. Mapping every SLA to its supporting OLAs and contracts is the unglamorous audit that separates promises from wishes — and it's the four-dimensions model (dimension three: partners and suppliers) earning its keep.

Vocabulary that prevents real arguments

Response time is when a human meaningfully engages with the ticket; resolution time is when service is restored. Users routinely hear the first number and expect the second — a 15-minute response promise is not a 15-minute fix promise, and every service desk veteran has had that conversation. Availability percentages deserve equal suspicion: 99.9% sounds like perfection and permits roughly 8.7 hours of downtime a year (or ~43 minutes monthly). Whether that's measured across all hours or business hours changes the promise enormously. Precision here isn't pedantry; it's the difference between an agreement and a future dispute.

Trail note

The trap with the memorable name: the watermelon SLA — green on the outside, red on the inside. Every metric on the dashboard is met, yet customers are furious: tickets closed within target but reopened three times; responses on time but useless; availability technically achieved while the service crawled. It happens because metrics measure what's countable, not what's valuable — and people optimise what's measured. ITIL 4's correction is watermelon-proofing: pair the operational numbers with experience measures (satisfaction, outcome achieved) so the dashboard can't be green while the humans are red. When you hear "XLAs" — experience level agreements — this is the idea being sold.

The craft, distilled: promise slightly less than you can deliver, measure what customers actually feel, and check the promises underneath your promises. Organisations that do this look boring and reliable. That's the compliment.

Next waypoint — Summit

The CMDB: IT's Map of What Exists and What Depends on What →