SLA (Service Level Agreement)

An SLA proves its worth the day an incident happens: if the application goes down on a Friday evening, the SLA states how fast the provider must respond and how quickly the issue must be resolved, depending on whether it's a blocking bug or a minor defect. For a company outsourcing maintenance or augmenting its team with an external provider, this document is what lets them check, in writing, that commitments are actually being met month after month.

A poorly negotiated SLA often boils down to an impressive uptime figure (99.9%) without detailing response times by severity level, making it useless when a real problem hits. It's also worth checking what penalties apply if the SLA isn't met, an SLA with no real consequence doesn't commit anyone to anything. A simple SLA with two or three clearly defined severity levels beats a long but vague document.