A customer service SLA, or service-level agreement, defines the support a customer or business can expect and how performance will be measured. For outsourced customer support, it should also explain who owns missed targets, which exceptions apply and what happens when the agreed service is not delivered.
The weak version says “respond quickly and provide excellent service”. The useful version states the channel, priority, operating hours, clock rules, evidence and owner. This article is an operational drafting template. Your procurement and legal teams should turn the agreed service model into the appropriate contract.
Start with the service promise, not the target number
An SLA cannot rescue an unclear scope. Before setting a response target, decide what the team will handle, which customers it serves, when it operates and what it is authorised to resolve. A team that only records an enquiry cannot honestly promise the same outcome as a team authorised to complete the request.
Separate a supplier agreement from a customer-facing promise. Your provider may have internal response and escalation commitments that help you meet a different promise to the end customer. Document the relationship so neither side assumes that one replaces the other.
Use our outsourced customer-service guide for the broader operating model. This page focuses on the service definition and measurement needed once you know what you are buying.
The customer service SLA template
Download the customer service SLA worksheet (text) and use it as the operational brief for your service schedule. It includes metric definitions, ownership and pre-launch test cases.
Copy these fields into your service schedule and complete them with the people who will run the work. Leave a field visibly unresolved rather than filling it with a target nobody has tested.
Scope and coverage
- Services included: channels, contact reasons, customer groups and tasks.
- Services excluded: specialist decisions, sensitive exceptions and work retained internally.
- Operating calendar: opening hours, time zone, weekends, public holidays and daylight-saving arrangements.
- Dependencies: systems, knowledge, client approvals and third-party actions the team needs.
- Accountable owners: the client's service owner and the provider's operational lead.
For a US business using teams in different locations, “business hours” is not a complete definition. Name the time zone and calendar. When after-hours support is outsourced, state whether a case is handled, triaged or only acknowledged overnight.
Priority definitions
Define priority through customer impact and urgency, with examples. Explain who can change a priority and record the reason. If every customer or agent can mark an enquiry urgent without review, the priority system stops helping anyone.
For each priority, write the first-response objective, the next-update requirement, the resolution objective where appropriate and the escalation route. A standard order-status enquiry, a widespread service interruption and an account-security issue may require different responses.
Measurement rules
For every metric, complete the same record:
Metric name; eligible contacts; clock start; clock stop; pause rules; operating calendar; target; reporting period; source system; owner; remedy if missed.
This definition is as important as the target. For example, a first response should specify whether an automated acknowledgement counts. A resolution measure should explain what happens when a ticket is reopened. Exclusions need a documented reason and an owner, otherwise they become a way to remove inconvenient cases.
Choose the right service measures
First response time
Measure the time until the first eligible reply or action under the agreed definition. It indicates access to support, not whether the problem was solved. Report the proportion meeting the target and the age of outstanding cases, rather than relying only on an average that can hide long waits.
Resolution and follow-up
Define what resolved means for each relevant contact type. Some cases require another department or an external supplier. Keep the customer informed during those dependencies and retain visibility of the total customer wait, even where a contractual clock legitimately pauses.
Separate an update commitment from a resolution commitment. A useful update explains what has happened and what is next. Repeated messages saying “we are looking into it” may meet a poorly written update rule while adding little value.
Voice service level
For calls, define the offered population, answer-time threshold, reporting interval and treatment of short abandons, callbacks and transfers. A monthly average can conceal repeated failures in the same busy intervals. Align the target with a workforce plan that can actually staff those intervals.
Quality and critical failures
Add a quality-assurance scorecard with agreed critical-error rules and calibration. A fast, inaccurate answer should not be classed as a successful interaction simply because it beat the response target.
Backlog and repeat demand
Report unresolved work by age and priority, with reopen and repeat-contact measures where reliable matching is possible. These measures expose the work that can disappear behind a good first-response result.
Zendesk's explanation of SLA policies distinguishes response and resolution measures. Its developer documentation also shows why business-hour settings and priority-specific targets matter. Your operating definition should be agreed before someone configures those fields in a tool.
Test the target with real cases
Choose examples that cross a closing time, a weekend, a holiday and a handoff between teams. Work through when each clock starts, pauses and stops. Confirm that the dashboard reproduces the agreed answer.
Suppose an enquiry arrives just before the covered shift ends and the customer replies after closing. The agreement needs to settle whether the follow-up clock runs overnight, restarts at opening or follows a separate after-hours queue. There is no useful answer until the coverage promise is defined.
Test reopened cases too. A team should not improve its reported resolution time by closing incomplete tickets and treating the customer's return as an unrelated new contact. Keep an audit trail of status and ownership changes.
Include client responsibilities
The provider cannot reliably resolve a refund if the client has not granted the necessary authority or supplied an approval route. The agreement should state who maintains knowledge, authorises exceptions, provides access and responds to escalations.
Use a responsibility record with a named owner, expected response, backup contact and escalation path for each dependency. Report supplier-controlled issues separately from client-controlled blockers, while keeping the complete customer journey visible.
This is where contact centre consulting can help: test the operating assumptions, evidence and accountability before a contract turns them into a recurring argument.
Agree the review and recovery process
Define when performance is reviewed and what evidence both sides receive. For a missed target, record impact, cause, immediate containment, corrective action, owner and review date. Check whether the action worked, rather than closing it because somebody delivered a presentation.
Agree how service credits, remedies or other commercial consequences work with the people responsible for the contract. The operational aim remains customer recovery and prevention of repeat failure. A credit does not automatically remove a backlog or repair an incorrect answer.
Include a change process. New channels, launches, policy changes and demand outside the agreed forecast can change the staffing and cost model. Record the assumptions, the threshold for review and who approves a change. Our implementation process connects these decisions with recruitment, training and launch readiness.
Before you accept an outsourcing proposal
Ask the provider to demonstrate the reporting definition using a sample case and to explain how its staffing plan supports the proposed targets. Confirm that the quoted rate includes the leadership, quality and reporting capacity the agreement requires.
Then ask who will act when the dashboard turns red outside normal office hours. If the answer depends on a manager who is not on duty and has no backup, the escalation promise is incomplete.
For a support model designed around your hours, channels and customer needs, explore customer service outsourcing. A useful SLA should describe that model accurately, not make an unsuitable model look reassuring.
Frequently asked questions
What is the difference between an SLA and a KPI?
A key performance indicator monitors an aspect of performance. An SLA records an agreed service commitment and how it is assessed. Some KPIs support the SLA without becoming contractual commitments themselves.
What is a good customer service response-time target?
It depends on the channel, urgency, customer promise and staffed coverage. Test the proposed target against arrival patterns and capacity. Do not copy a number from another business without checking those conditions.
Should an automated reply count as a first response?
Only if the agreement explicitly defines it that way. Separate acknowledgement from a meaningful response where customers need a person or a substantive action. Otherwise a fast acknowledgement can hide a long wait for help.
Can we use one SLA for phone, email and chat?
You can keep one service schedule, but the measures and clock rules need to match each channel. A live conversation and an asynchronous email queue do not create the same waiting experience or staffing requirement.
Who owns an SLA in an outsourced customer-service team?
Name an accountable owner on both sides. The provider owns delivery within its scope, while the client retains responsibility for its customer promise, policy decisions and agreed dependencies.
About the author
Oliver Catt is the founder of The Cattalyst. He has spent 25 years inside contact centres, customer service, sales and training operations, and now helps organisations develop their leaders and build outsourced teams through a global partner network.
Start a conversation with Oliver
