SLA credits explained: what you get when a provider misses its uptime target
How SLA service credits work, how providers measure downtime, why you must claim them, and how to keep your own evidence so the claim isn't denied.
· 7 min read · By the Spot Downtime team
An SLA (service level agreement) is a promise about availability, and a service credit is what the provider gives you when it breaks that promise. Credits are usually smaller than people expect, they rarely arrive without a request from you, and the way downtime is measured often decides the outcome. If you depend on a hosting, cloud or API provider, it pays to know the mechanics before you need them.
How a credit works
Most SLAs define a monthly uptime percentage and a ladder of credits. If the month's uptime falls below the commitment, you get a percentage of that month's fee back as a credit on a future invoice. The structure is typically like this (a hypothetical example, not any specific provider's terms):
| Monthly uptime | Credit |
|---|---|
| 99.9% or higher | None |
| Below 99.9%, at or above 99.0% | 10% of the monthly fee |
| Below 99.0%, at or above 95.0% | 25% of the monthly fee |
| Below 95.0% | 100% of the monthly fee |
Things that are true of almost every SLA, whatever the numbers:
- Credits aren't refunds. They offset future charges, and often expire or can't be paid out as cash.
- They're capped at the fee for the affected service, not at what your outage cost you.
- They apply to a specific service, not the whole account. A database outage may not count against the compute SLA.
- You usually have to ask, within a claim window that can be as short as a month or two after the incident.
That means an SLA is better understood as a statement of how seriously the provider takes reliability than as insurance. If a single outage costs you more than a month of fees, the credit won't cover it.
What counts as downtime
This is where claims succeed or fail. Read the definition, not just the percentage. Look for:
- How an outage is detected. Many SLAs measure on the provider's side, from its own health checks, and count only requests that failed on its network. Your users may have seen errors the provider doesn't count.
- Minimum duration. Some only count unavailability lasting several consecutive minutes.
- Exclusions. Scheduled maintenance, force majeure, problems caused by your own configuration, and failures of third parties are commonly carved out.
- Error-rate definitions. Some define downtime as an error rate over a threshold in a time window, not a full outage. A half-broken service may not count.
- Scope. A single region, zone or instance being down may not qualify if the service is defined as the whole multi-zone deployment.
What each target allows
To judge a commitment, translate it to time. Over a 30-day month, 99.9% allows about 43 minutes of downtime and 99.99% about 4 minutes. The full table goes from 99% to 99.999%, and the 99.99% guide covers when that target is realistic.
Uptime calculatorConvert an uptime percentage into allowed downtime per day, month and year, or the reverse.Your real uptime is lower than any single SLA
If your service depends on several providers and any one of them failing takes you down, the availabilities multiply. Two dependencies at 99.9% each give 0.999 × 0.999 = 99.8%, which allows around 86 minutes a month, double what either promises. Add a third and it gets worse. A provider's credit for its 40 minutes of downtime says nothing about the lost sales on your side.
Don't promise customers more than you can deliver
How to claim a credit
- Read the claim process. Find the deadline, what the request must include and where to submit it (usually a support ticket).
- Write down the exact times, with time zones: when the problem started, when it ended and which resources were affected.
- Attach your own evidence. Logs, request IDs and monitoring records from outside the provider's network carry more weight than "it felt down".
- Calculate the uptime yourself using the provider's formula, and state which credit tier applies.
- Keep the reply, and check the credit actually appears on the next invoice.
Why independent evidence matters
If the only record of an outage is the provider's own status page, the provider decides whether it happened and for how long. A record taken from outside, with timestamps and the reason for each failure, gives you something to point to. Spot Downtime's incident history logs each outage from the first confirmed failure to recovery, with its duration and reason, and shows uptime for 24 hours, 7 days and 30 days per monitor. It's evidence from outside the provider, though what counts for a claim is still defined in the contract.
When you offer an SLA yourself
The same mechanics apply in reverse. Define downtime precisely, say how you measure it, list the exclusions, and publish the numbers where customers can see them. A public status page with uptime history makes the measurement transparent, and the practices in status page best practices help keep it credible. Offer a target you can meet with margin: a missed SLA costs credits and trust, and a modest promise you keep is worth more than an ambitious one you don't.
Get new guides by email
Practical posts on monitoring, outages, error fixes and on-call, when they're published. No spam.
Keep reading
- Uptime · SLAs · AlertsError budgets for small teams: a practical guideAn error budget turns an uptime target into minutes you can spend. How to calculate it, decide what counts, and use burn-rate alerts without a dedicated SRE team.November 11, 2026 · 7 min read
- Uptime · SLAs99.99% uptime: how much downtime is that?Four nines allows 52 minutes of downtime a year, or 4 minutes 23 seconds a month. What that budget means for deploys, on-call and monitoring, and when 99.99% is worth promising.October 27, 2026 · 5 min read