Service level agreement (SLA)

A service level agreement (SLA) is a documented commitment between a service provider and a customer, defining the level of service the customer can expect. In customer support, SLAs typically specify how fast tickets must be answered, how long resolutions can take, and what happens if those targets are missed.

Short version: an SLA is a promise about service quality, with numbers attached.

What's in a customer service SLA

A solid customer service SLA usually defines:

• Response time targets. How quickly the team will reply, often broken down by channel (email, chat, phone) and priority.
• Resolution time targets. How long the team has to fully resolve different categories of issues.
• Hours of operation. When the SLA applies. Many SLAs run only during business hours, not 24/7.
• Priority definitions. What counts as a P1, P2, or P3 ticket. Without this, the rest of the SLA is meaningless.
• Escalation paths. What happens when an issue blows past its target.
• Reporting cadence. How and when SLA performance is shared with the customer.
• Consequences. What happens if SLAs are missed. Service credits, fee reductions, or other remediation.

Internal vs external SLAs

Two flavors, with different uses:

• External SLAs are commitments made to customers, often as part of a contract. Common in B2B and enterprise relationships. Missing one usually has financial or contractual consequences.
• Internal SLAs (sometimes called OLAs, operational level agreements) are commitments between teams within the same company. Support might have an internal SLA with engineering for bug response time, for example.

Most B2C ecommerce companies don't publish formal external SLAs to customers, but most run with internal SLA targets the support team is measured on. Same idea, different audience.

SLA vs SLO vs SLI

Three terms that get mixed up:

• SLA (Service Level Agreement): The contract. The promise to the customer.
• SLO (Service Level Objective): The internal target. Usually stricter than the SLA, to give the team a buffer.
• SLI (Service Level Indicator): The actual measurement. The number you report against the objective and the agreement.

A simple example: the SLA says first response within 24 hours. The SLO says within 4 hours (so the team has buffer). The SLI is the actual measured average, say 2.5 hours.

Why SLAs matter in customer service

Three reasons:

• They set customer expectations. Without an SLA, customers invent their own expectations, and they're usually unrealistic.
• They drive team focus. A team without targets prioritizes whichever ticket feels loudest. A team with targets prioritizes the ticket closest to breaching.
• They make performance visible. SLA reports are the easiest way to show whether the support function is delivering or slipping.

Common metrics tracked against SLAs

The metrics that typically anchor a customer service SLA:

• First response time (FRT): How fast the first reply goes out.
• Resolution time: Time from open to close.
• SLA attainment: Percentage of tickets that hit their target. The headline number on most reports.
• Escalation rate: How often tickets get bumped to higher tiers, often correlated with at-risk SLAs.
• Backlog: Tickets currently open. A growing backlog is usually the leading indicator of SLA misses.

How AI changes SLAs

AI in customer support doesn't just help meet existing SLAs. It often changes what's reasonable to commit to.

Three patterns we see across Neople customers:

• FRT targets shrink dramatically. Tickets that AI handles directly are answered in seconds, not hours. SLA promises that used to read "first response within 24 hours" can move to "within 1 hour" without adding people.
• Resolution times drop. AI surfaces the right context, drafts the right answer, and lets human agents close tickets faster on the work that does need a human.
• SLA attainment improves. Across our customer base, SLA attainment typically rises by 15 to 25 percentage points within the first quarter of deployment. Same SLA, more met.

Common mistakes

• Setting unrealistic SLAs to win deals. Missing them costs more than not promising them.
• Confusing SLA, SLO, and SLI. The agreement is not the objective. The objective is not the indicator.
• Tracking SLA attainment without segmenting. A 95% attainment rate hides whether the 5% being missed are minor or critical.
• Treating the SLA as static. Customer expectations and operational capacity both change. Review SLAs at least yearly.