What is service level agreement?

Why it matters

An SLA you cannot keep is worse than no SLA, because it converts a slow reply into a broken promise. Most breaches happen at predictable times, overnight, at weekends, and during spikes, which means most SLA problems are coverage problems rather than performance problems.

What to know

• Separate response SLAs from resolution SLAs. Committing to resolution times on issues you do not control is a trap.
• State the clock. Business hours or round the clock changes the promise completely and is the most common ambiguity.
• Breaches cluster outside working hours, so coverage design determines compliance more than team speed does.
• Early-warning alerts before the deadline are what actually prevent breaches. Reporting after the fact does not.
• Tiered SLAs by customer value are standard and should be encoded as routing rules, not left to judgement.

An example

Example: a contract promises a four-hour first response with no clock stated. A ticket arrives at nine in the evening. The customer counts four hours, the vendor counts business hours, and both believe they are right.

Common mistakes

• Promising resolution times for issues dependent on third parties
• Failing to state whether the clock runs outside business hours
• Discovering breaches from reports rather than preventing them with alerts
• Setting one SLA across all customer tiers and issue severities

Questions