Sit in enough support planning meetings and you will hear "SLA," "KPI," and "OLA" used as if they were roughly the same thing — a number the team is supposed to hit. They are not the same thing, and the difference matters more once a support org has more than one team involved in resolving a ticket.

SLA: the promise to the customer

A service-level agreement (SLA) is the commitment your company makes to the person outside it — usually a first-response or resolution time target. It is external-facing by nature, even when the customer never sees the exact number: it is the standard you are holding yourself to on their behalf.

OLA: the promise between internal teams

An operational-level agreement (OLA) is the internal version of the same idea — the time the billing team commits to respond to a support agent's escalation, for instance, so the support agent can in turn hit the SLA they promised the customer. If a ticket depends on three internal handoffs, the SLA is only achievable if every OLA behind it is realistic and actually met.

KPI: how you know either one is working

A key performance indicator is a measurement, not a promise — CSAT, average handle time, ticket volume per agent. KPIs tell you whether your SLAs and OLAs are sustainable or whether the team is hitting the target by cutting corners that will show up somewhere else, like a CSAT score quietly declining while resolution time looks great.

Why the distinction actually matters

Treating all three as interchangeable usually shows up as one specific failure: a customer-facing SLA that depends on an internal team with no OLA at all, so the promise to the customer rests on an assumption nobody actually agreed to. Naming the layers separately is what makes it possible to find that gap before a customer does.

Building this in practice

Our guide to building an SLA program that works goes deeper on setting realistic targets per channel and designing escalation that fires before a breach, not after.