October 13, 2026
A service level agreement (SLA) is the part of a customer contract that promises measurable service levels, usually uptime and support response times, and says what the customer gets when you miss them. Most startups don’t need one until an enterprise customer asks. When one does, promise only what you can measure and reliably hit, make service credits the customer’s only remedy, cap them, and exclude what you can’t control.
Key takeaways
Usually when a larger customer asks for one, often around the first sizable annual contract. Until then, keep service commitments out of your contracts. An SLA turns “we’ll try our best” into a promise with a price attached, so offer it when the deal justifies it, and consider offering a stronger SLA only on higher-priced plans.
A good SLA is short and precise. It describes the service covered, defines each metric and how it is measured (for example, uptime measured by your monitoring tool every few minutes over a calendar month), sets support response times by severity, explains how customers report issues, and states the credit schedule, the claim process and the exclusions. Broader terms such as data security, liability caps, governing law and dispute resolution belong in the main agreement, not the SLA. See How Do I Negotiate Customer Contracts as a Startup?
Pick targets you already beat in practice, with room to spare. Over a 30-day month, 99.5% uptime allows about 3.6 hours of downtime, 99.9% about 43 minutes, 99.95% about 22 minutes and 99.99% about 4 minutes. 99.9% is a common enterprise ask. Set your internal target higher than your contractual one so an ordinary bad month doesn’t trigger credits.
Response time commitments should be about responding, not resolving. “We will respond to a Severity 1 issue within four hours” is a promise you control; “we will fix it within 24 hours” may not be.
Credits are a percentage of the monthly fee for the affected service, applied against future invoices. Use tiers and a cap, not a per-increment formula. A formula like “5% credit for every 0.01% below target” can owe the customer more than the monthly fee after a short outage. A typical capped schedule for a 99.9% commitment looks like this:
| Monthly uptime | Service credit |
|---|---|
| 99.0% to below 99.9% | 10% of monthly fee |
| 95.0% to below 99.0% | 25% of monthly fee |
| Below 95.0% | 50% of monthly fee (maximum) |
The contract should also say that credits are the customer’s sole and exclusive remedy for SLA failures, that the customer must request a credit in writing within a set period (often 30 days after the month ends), and that credits have no cash value. Enterprise customers often ask for a right to terminate after repeated failures, for example missing the SLA in three months out of any six. That can be a reasonable compromise if it replaces larger damages claims.
Exclude scheduled maintenance announced in advance, emergency maintenance within reasonable limits, force majeure events, outages of third-party providers outside your control (such as your cloud host), problems caused by the customer’s own systems or misuse, and beta or free features. Without these exclusions, you are guaranteeing other companies’ performance as well as your own. See What Should I Check Before Signing a Vendor or SaaS Contract?
For more on contracts, see How Do I Draft a Strong Contract for My Startup? and What Are the 10 Contracts Every Startup Should Have on Hand? If you’d like help with a customer agreement or SLA, see our Contract Review service or contact us.
