This post is for information only. You are responsible for reviewing and using this information appropriately. This content doesn’t contain and isn’t meant to provide legal, tax, or business advice. Requirements are updated frequently and you should make sure to do your own research and reach out to professional legal, tax, and business advisers, as needed. To sell products using the Shopify platform, you must comply with the laws of the jurisdiction of your business and your customers, the Shopify Terms of Service, the Shopify Acceptable Use Policy, and any other applicable policies.
A service-level agreement (SLA) sets the standards a provider agrees to meet and the remedies available if service falls short. For an online store, those terms cover tools that process orders or handle customer inquiries.
An outage can interrupt sales and leave customers waiting. Uptime Intelligence’s 2026 outage analysis found that 57% of data center operators responding to its 2025 survey said their most recent major outage cost more than $100,000.
This guide covers how SLAs work, what their terms cover, and how to assess an agreement for your business.
What is a service-level agreement (SLA)?
A service-level agreement (SLA) is an agreement between a service provider and its customer that defines expected performance and each party’s responsibilities. It also explains what happens if the provider falls short of those expectations.
SLAs cover services ranging from software to order fulfillment, and they can also govern services between departments within a business. For example, a customer service platform could commit to 99.9% monthly uptime. Its SLA would explain how availability is measured, which disruptions count as downtime, and the remedy available if performance falls below that threshold.
Software SLA remedies typically take the form of service credits applied to future bills, rather than automatic refunds. Customers generally have to submit a claim with evidence of the disruption by a set deadline.
SLA coverage varies by subscription and contract. An agreement can come with your software subscription or through a separate paid support plan. For example, cloud platform Momento’s Enterprise support plan offers a 99.99% service SLA. The plan carries a one-year commitment and lists pricing of $5,000 a month or 15% of eligible infrastructure spending.
Why do SLAs matter?
An SLA gives you specific standards for evaluating a provider’s performance. For example, a support agreement could promise an initial response to urgent issues within one hour. Written commitments explain what the provider must deliver and the consequences for missing a target.
The agreement also assigns responsibilities to both parties. The provider commits to the stated service level, and the customer follows the reporting procedures and meets any conditions for coverage.
Offer protection and recourse
On May 27, 2026, Slack experienced a global disruption that affected messaging and automated workflows. Slack restored services in stages, with the final issues resolved on May 28.
For a covered disruption, an SLA spells out the remedy you can claim. Atlassian’s SLA, for example, offers credits against future payments when eligible products miss their uptime commitments. Customers must submit a claim within 15 days after the affected calendar month ends.
These credits recover part of your service costs. Their value reflects the service fee, so lost sales and staff time can exceed the amount credited.
Types of SLAs
SLAs can organize commitments around a customer, a service, or several layers of an agreement. The commitments vary by the work involved:
| Service category | Common commitments | What to check |
|---|---|---|
| Software/SaaS | Platform availability and processing performance | Covered features, measurement periods, and maintenance exclusions |
| Customer support platforms | Platform uptime and vendor support response times | Separate the vendor’s commitments from response targets you set for your own agents |
| Vendors/suppliers | Delivery deadlines and acceptable defect rates | How late or defective deliveries are recorded and remedied |
| Fulfillment/shipping | Order accuracy and dispatch or delivery deadlines | Order cutoffs and where responsibility passes from warehouse to carrier |
For supplier agreements, your vendor management records should capture the delivery and quality targets, and each supplier’s results.
Customer-based SLA
A customer-based SLA covers the services provided to a particular customer, whether that customer is an outside business or an internal team.
The agreement sets commitments around that customer’s needs. For example, a retailer could negotiate one agreement covering cloud storage and technical support.
Service-based SLA
A service-based SLA sets a standard commitment for customers using a particular service. Customers on the same plan receive the same terms, but commitments can differ between plans.
For example, Atlassian’s SLA sets a 99.9% monthly uptime commitment for covered Jira functions on Premium and 99.95% on Enterprise.
Internal SLA
An internal SLA is a form of customer-based SLA between teams within the same business. The team receiving the service is the internal customer.
At a small online store, the inventory manager could agree to investigate stock discrepancies reported by customer support within two business hours. Unresolved discrepancies would then go to the owner.
An external customer SLA covers work assigned to an outside provider. A store can have both agreements, with separate commitments for its staff and the provider.
Multi-level SLA
A multilevel SLA organizes commitments into three parts:
- Corporate: Terms shared across customers and services, such as a standard incident-reporting procedure.
- Customer: Terms for a particular customer or group, such as a retailer’s overnight support coverage.
- Service: Targets for a specific service, such as monthly uptime for hosted checkout.
The corporate part establishes common terms. The customer and service sections add requirements for the account and the work being delivered.
Key components of service-level agreements
Review an SLA together with the main service contract and any separate security terms. Check how each commitment will be measured and what you must do to claim a remedy.
| Component | What it specifies | What buyers should check before signing |
|---|---|---|
| Service description | Services and features covered by the agreement | Confirm coverage for the functions your business relies on |
| Performance metrics | Measurable targets and the period assessed | Check the calculation method and definitions of downtime or failure |
| Reporting and monitoring | Records available to verify performance and completed fixes | Confirm access to incident records and the evidence accepted in disputes |
| Responsibilities | Tasks assigned to the provider and customer | Identify customer actions or configurations needed for coverage |
| Penalties and remedies | Credit percentages linked to missed performance thresholds | Check eligible charges, credit caps, claim deadlines, and required evidence |
| Exclusions and limitations | Events or conditions excluded from coverage | Review maintenance allowances and responsibility for third-party failures |
| Escalation paths | Who takes over an unresolved issue and when | Confirm escalation contacts and coverage outside business hours |
| Security and breach notification | Security duties and incident-notification requirements | Check what triggers notification, its deadline, and who receives it |
Service credits are calculated as a percentage of eligible charges for the affected service, with larger credits for lower performance. For example, Amazon EC2’s regional SLA offers a 10% credit when monthly uptime falls below 99.99% but stays at or above 99%. If eligible charges total $500, that credit is $50.
For monitoring, compare the provider’s reports with your incident records and request evidence that corrective work succeeded. The FTC’s data breach response guide specifically tells businesses to verify that providers have fixed vulnerabilities they claim to have remedied.
Shopify’s app support process shows how escalation works. For eligible App Store apps, go to Settings > Apps, open the app’s menu, and select Get support. The request goes to the developer. Shopify follows up after three days and contacts the developer again if your issue is unresolved.
Shopify reports 99.9% platform uptime across major services and geographic regions over the previous 90 days, subject to stated exclusions.
What SLA metrics mean: uptime, response time, and resolution time
Uptime measures availability over a defined period. For a time-based SLA, calculate permitted downtime by multiplying the measurement period by the unavailable percentage. A 99.9% target allows 0.1% downtime under that calculation.
| Uptime commitment | Downtime within the target in a 30-day month |
|---|---|
| 99.5% | 3 hours, 36 minutes |
| 99.9% | 43 minutes, 12 seconds |
| 99.95% | 21 minutes, 36 seconds |
| 99.99% | About 4 minutes, 19 seconds |
These calculations assume continuous measurement across all 43,200 minutes in the month. The SLA’s definitions and exclusions determine which disruptions count. For a hosting SLA, check whether planned website maintenance counts toward downtime and how much notice the provider must give.
A single 40-minute outage could fit within a 99.9% monthly target and still take checkout offline during a Black Friday promotion. The same outage on an average day could affect far fewer shoppers. The percentage measures availability, while the timing determines how much business is interrupted.
Configuration can change the commitment. Amazon EC2’s SLA sets a 99.5% target for an individual instance. Its 99.99% regional commitment applies to qualifying deployments running across multiple Availability Zones, which are separate infrastructure locations within a region.
If an app stops importing orders, support could meet a 15-minute response commitment by replying to your ticket while the app is still offline. A four-hour resolution commitment would set a separate deadline for fixing the problem. Check when each clock starts and whether it pauses outside business hours or while support waits for your reply.
When comparing paid support plans, check how each would handle the same issue. For a normal-priority ticket, GitHub’s support SLA promises an initial response within 48 hours on Premium and 24 hours on Premium Plus. GitHub starts that clock when it assigns the ticket’s priority. Any wait before that assignment adds to the time you spend waiting for a response.
SLA best practices
Review an SLA against the work your business needs the provider to perform. Check how you would report a failure, claim a remedy, and leave the service if problems persist.
Evaluate SLA terms before commitment
Check that performance targets have straightforward definitions and measurement periods. For a response-time commitment, confirm when the clock starts and which support hours count. Review the remedy for each missed target, how credits are calculated, and the deadline for submitting a claim.
Other terms to consider:
- Check cancellation notice periods and any early termination fees.
- Confirm which data you can export, its format, and how long you retain access after cancellation.
Escalation clauses also need named contacts and deadlines for passing unresolved issues to someone with authority to act.
For cloud services covered by the EU Data Act, providers must remove obstacles to switching. Covered SaaS providers must make open interfaces available and export data in a commonly used, machine-readable format. Check that the contract explains how the provider will handle your exit.
Compare SLAs across multiple providers
Put covered services and performance calculations side by side, then compare exclusions and remedies.
For example, Intercom’s SLA sets a 99.8% monthly availability target for covered services. A verified failure in two consecutive months gives customers a right to terminate the service and receive a refund of unused prepaid fees, subject to its notice deadline.
Zendesk offers service credits to qualifying Premier customers when covered products fall below its 99.9% monthly availability commitment. The remedy and eligibility rules therefore deserve as much attention as the percentage.
If a provider’s SLA is unavailable online, request the terms that would apply to your account.
Seek legal counsel for SLA review
An SLA can be contractually enforceable when its obligations form part of a valid agreement accepted by both parties. A lawyer can check whether the terms are clear and enforceable under the governing law.
Ask counsel to review limits on remedies and liability. Software SLAs designate service credits as the exclusive remedy for missed availability commitments. For example, Atlassian limits that remedy to credits and caps the credit applied to an invoice at the amount charged for the affected product.
For a service that handles essential operations, ask counsel whether repeated failures give you the right to terminate and recover unused prepaid fees.
Document and report service failures
Record when a disruption begins and ends, including the time zone. Identify the affected feature and describe what failed. Save error messages or screenshots alongside support tickets and the provider’s incident updates.
Monitor both the platform and your store. Shopify Status publishes incidents affecting Shopify services and links to store-specific status information. The third-party Uptime app also has monitoring and automated store tests on eligible plans.
These tests can detect problems such as a broken add-to-cart flow and trigger alerts. Check the plan’s testing features before installing it.
Submit claims through the channel specified in your SLA and retain confirmation of submission. Check which evidence the provider accepts and whether the deadline runs from the incident or the end of the billing month. Assign one person to track claims through to the resulting credit.
Review your SLA at renewal
A move to a cheaper plan could replace round-the-clock support with business-hours coverage. A provider acquisition or a new hosting partner can also change who handles an outage.
Your contract management records should include the current SLA and its renewal notice deadline, so you can compare proposed pricing and service terms before committing. Revised uptime calculations or new exclusions could reduce your eligibility for credits.
If previous claims stalled because you and the provider disagreed about downtime, ask for an agreed measurement method and a process for resolving disputes in the renewed contract.
How to write a service-level agreement
Replace each bracketed field with agreed terms. Keep the SLA consistent with the main service contract and follow this template:
- Scope. [Provider] will deliver [services/features] to [customer] from [date] for [term]. This SLA forms part of [main agreement].
- Performance metrics. Uptime will be [X%] per [period], calculated as [formula] from [measurement source]. For [severity], initial response is due within [time] and resolution within [time], measured during [support hours] under [clock-start and pause rules].
- Responsibilities. Provider will [duties]. Customer will [duties], including [required configuration or access].
- Reporting. Provider will deliver [report] every [interval]. Both parties will retain [records] for [period] and resolve measurement disputes through [process].
- Escalation. Report incidents through [channel]. Escalate unresolved [severity] incidents to [role/contact] after [time], with updates every [interval].
- Remedies. For [missed threshold], provider will credit [percentage] of [eligible fees], capped at [amount]. Claims need [evidence] by [deadline]. Other remedies and limits are [terms].
- Exclusions. Excluded events are [events]. Planned maintenance requires [notice] and is limited to [window/duration].
- Security and notification. Provider will maintain [controls] and notify [contact] within [time] of [defined incident trigger], with updates every [interval].
- Exit terms. Termination requires [notice], subject to [early-exit rights/fees]. Provider will export [data] in [format] within [time], assist with [migration], and delete data under [retention terms].
The European Commission’s non-binding cloud-contract clauses include optional structural references for switching and termination provisions, along with security and liability terms. Businesses can adapt these voluntary clauses to their agreements.
Service-level agreement FAQ
What are the three types of SLAs?
The three main types of service-level agreements are:
- Customer-based agreements cover services for a particular customer.
- Service-based agreements set shared commitments for customers using a specific service.
- Multilevel agreements organize terms into corporate, customer, and service layers.
What is an example of an SLA?
Atlassian’s SLA commits to 99.9% monthly uptime for eligible Premium cloud products and 99.95% for Enterprise. Eligible customers can claim service credits when covered products miss those targets. Customers must submit claims within 15 days after the affected calendar month ends.
What is the purpose of an SLA?
An SLA defines the service a provider commits to deliver and how performance will be measured. It assigns responsibilities to both parties and specifies what happens when agreed standards are missed, including how customers report problems and claim remedies.
What is the difference between an SLA and KPI?
An SLA defines agreed service standards and the consequences of missing them. A key performance indicator (KPI) measures performance against a goal. For example, monthly uptime can be a KPI, while an SLA sets the minimum uptime commitment and the remedy for falling below it.
Is an SLA legally binding?
An SLA can be legally binding when it meets the requirements of a valid contract or is incorporated into one. Enforceability depends on its wording and governing law. Software SLAs often limit remedies to service credits, so have a qualified lawyer review the obligations and liability limits.
