A service level agreement (SLA) is the part of a service contract that turns “we’ll keep it running” into numbers: how available the service will be, how fast the provider responds when something breaks, how that’s measured, and what the customer gets back when targets are missed. IT providers, hosting and software companies, managed service providers and outsourced support desks all sign them. This free SLA template is written to attach to a main services contract.

What does an SLA actually commit a provider to?

It commits the provider to measurable targets, not to perfection. NIST’s glossary, quoting its interconnection guidance, describes an SLA as a commitment between a service provider and one or more customers covering specific aspects of the service, such as responsibilities, the expected performance level (reliability, acceptable quality, response times), and requirements for reporting, resolution and termination. That’s a good checklist for what yours should include.

The key word is measurable. “Fast support” can’t be enforced. “First response to a Severity 1 ticket within 30 minutes, 24/7” can.

Where does an SLA fit with your other contracts?

An SLA usually lives inside something bigger. The main agreement handles price, confidentiality, ownership, liability and termination, and the SLA handles performance. If you work with a client on repeat projects, put the SLA under a master service agreement. For a one-off job, a service agreement may be all you need, since there’s no ongoing service to measure. If your concern is protecting information the provider sees rather than uptime, an NDA is the document you want.

What should a service level agreement include?

Service description and hours

What’s covered, what isn’t, and when the clock runs. A customer-facing web app might be 24/7. A payroll help desk might be weekdays 8 to 6 Eastern. Be explicit about time zones.

Availability target and how it’s calculated

The uptime percentage and the formula. Our template measures availability monthly: total minutes in the month, minus excluded downtime, minus unplanned outage minutes, divided by total minutes minus excluded downtime. Pick a target you can actually hit. For a sense of scale, 99.9% allows about 43 minutes of downtime in a 30-day month; 99.5% allows about 3.6 hours.

Measurement

Who measures, with what, and how disputes about the numbers get resolved. Usually the provider’s monitoring counts, with the customer allowed to submit its own evidence. Define “unavailable” tightly, for example “the login page or API returns errors or doesn’t respond for five consecutive one-minute checks from two regions.”

Support severity levels and response times

A table of severity levels with a definition, a response target and an update frequency for each. Response time is not fix time. Most small providers can commit to responding quickly; far fewer can promise a fix within a set time, so the template separates the two.

Service credits

What the customer gets when availability falls short, usually a percentage of that month’s fee on a sliding scale, capped at some share of the monthly fee. The template requires the customer to claim credits within a set window and treats them as a price adjustment, not a penalty.

Exclusions

Downtime that doesn’t count: scheduled maintenance with notice, the customer’s own systems or misuse, third-party internet outages, force majeure events, and suspension for non-payment. Keep the list honest. An SLA with more exclusions than commitments isn’t worth signing.

Reporting and chronic failure

A monthly report, plus a way out: if the provider misses the target in, say, three months out of any six, the customer can terminate without penalty.

How would this work in practice?

Suppose a managed IT provider supports a 40-person accounting firm for $3,200 a month, with a 99.9% availability target for the firm’s hosted file server. In February, a failed storage update takes the server down for 95 minutes on a Tuesday afternoon. February has 28 days, or 40,320 minutes, so availability comes out at about 99.76%. Under the credit table in our template, that falls in the “below 99.9% but at least 99.5%” band, which earns a 10% credit, or $320 off March’s invoice, once the firm submits a claim. Nobody has to argue about what “reasonable uptime” means.

How do you fill it in and get it signed?

  1. Describe the service and the covered hours, including time zone.
  2. Set the availability target and credit bands that fit your price. Higher targets cost the provider more to meet.
  3. Write severity definitions that match real incidents you’ve seen, and set response targets your team can staff.
  4. List scheduled maintenance windows and notice periods.
  5. Attach the SLA to your main agreement and sign both together.

The federal ESIGN Act says a contract can’t be denied legal effect just because it’s in electronic form, so SLAs and their amendments can be e-signed like any other business contract. Our guide to getting a document signed by multiple people helps when both companies need more than one signer.

Download it, fill in the blanks, and send it for e-signature with any tool you like. (If you’d like to try SignWren when it launches, join the waitlist.)

This template and guide are general information, not legal advice. For a specific contract or dispute, talk to a lawyer licensed where you are.