Service level agreement: promising something you can actually meet

A service level agreement states what level of service a customer will receive, how it is measured, and what happens when it is not delivered. It is a contractual commitment rather than a marketing statement, and the difference matters: a target you publish becomes a promise somebody can hold you to.

The common failure is copying figures. A competitor promises a four-hour response, so you promise a four-hour response, without measuring what you currently do or arranging cover for the hours you are asleep. The first time somebody invokes it, you discover that the number was never grounded in anything.

What an SLA has to define

• The service in scope, and what is explicitly out. Vagueness here is where every dispute starts.
• The metrics, each with a definition. Response time means what precisely — first human reply, automated acknowledgement, or resolution? These are wildly different promises.
• The hours during which the clock runs. Four hours on a Sunday and four business hours are separate commitments; say which.
• Priority levels with definitions, so that severity is not argued about at the moment it matters.
• How measurement happens, and who reports it. An SLA neither party measures is decoration.
• Exclusions: customer-caused delays, third-party outages, planned maintenance, force majeure.
• Consequences of breach — service credits, escalation, termination rights — proportionate to the contract's size.
• The review cadence, because a number set two years ago rarely still fits.

Measure before you promise. Take your last fifty tickets or jobs, calculate what you actually achieved, and set the commitment somewhere you meet comfortably rather than at your best-case figure. An SLA met ninety-eight per cent of the time builds trust; the same number missed monthly destroys more goodwill than a slower promise ever would.

Response, resolution and uptime

Three commitments of very different difficulty. Response time — a human replying — is largely within your control and safe to promise. Resolution time depends on what broke, and committing to it means committing to problems you have not seen yet; a target rather than a guarantee is the honest form. Uptime is a percentage that sounds mild and is not: ninety-nine per cent allows over seven hours of downtime a month, while ninety-nine point nine allows around forty-three minutes. Know which one you are promising, because the operational cost between them is enormous.

SLA, OLA and the internal version

An SLA sits between you and a customer. An operational level agreement sits between internal teams and supports it — if you promise a customer four hours and your supplier's part of the chain takes a day, the external promise cannot hold. Underpinning contracts with third parties are the same idea one level further out. For a small business the useful version of this is simply checking that the promise you make outward is supported by the arrangements you have inward, which is where suppliers come in via vendor management and third party risk management.

Where it lives

Ettex Records holds the agreement, the priority definitions and the measurement record with review dates attached, so the commitment and the evidence sit together. The support side that has to deliver it is covered in help desk and support ticket, and satisfaction alongside it in customer satisfaction score.

Being direct: there is no SLA timer, no automatic breach alerting and no service credit calculation. Those are features of dedicated help desk products. What we hold is the document, the definitions and the record — and the honest observation that most small-business SLAs fail on definition rather than on tooling.