Most teams treat SLA credits as a rounding error. They are not. On any meaningful cloud spend, a single breach can be worth thousands of dollars, and the paperwork to collect it rarely takes more than an afternoon. The reason the money goes unclaimed is not that it is small. It is that nobody does the math.
This post does the math. If you want to plug in your own numbers, the SLA calculator on our homepage will turn your downtime into a dollar figure in about a minute.
What an SLA credit actually is
A Service Level Agreement is a promise about availability, usually stated as a monthly uptime percentage. When the provider misses that number for your resources, the contract entitles you to a service credit: a percentage of that service's monthly bill, applied to a future invoice.
The credit is not a refund of your losses. It does not cover lost revenue, engineering overtime, or churned customers. It is a fixed contractual remedy tied to one thing: how far below the target your measured uptime fell.
The tiers that decide the payout
Each provider publishes a table that maps uptime bands to credit percentages. The exact numbers vary by service, but the shape is consistent across the big three.
| Measured monthly uptime | Typical credit (compute-class SLA) |
|---|---|
| 99.0% to below 99.99% | 10% |
| 95.0% to below 99.0% | 25% |
| Below 95.0% | 100% |
Two things fall out of this table immediately. First, the credit is not linear. A little more downtime can jump you from a 10% tier to a 25% tier, which more than doubles the payout. Second, the "100%" tier is real: a bad enough month can wipe out that service's entire bill.
The dollar math on a real bill
Percentages are abstract, so let's attach money. Say a single service line (a compute fleet, a managed database, a load balancer tier) costs $40,000 a month.
- A 10% credit on that line is $4,000.
- A 25% credit is $10,000.
- A 100% credit is $40,000.
Now weigh that against effort. Confirming the breach, pulling evidence, and filing the claim is a one-hour to two-hour job. At $4,000 for the smallest tier, that is an effective rate most consultants would envy. At $10,000 it is difficult to justify skipping.
And these are per-service, per-incident figures. A provider region event that touches several of your services in the same billing month produces a separate credit calculation for each affected service. The totals compound fast.
Why "our spend is too small" is usually wrong
The instinct is to assume credits only matter at hyperscaler scale. Run the numbers on your own bill instead of guessing. Even a modest $8,000/month service line yields $800 at the 10% tier and $2,000 at 25%. That is real money for an hour of work, and it recurs every time the provider breaches.
The break-even is lower than almost anyone assumes. If your affected service costs more than a few hundred dollars a month, the credit is worth the paperwork. Our companion breakdown of the real cost of downtime walks through how these figures scale with spend.
Why the money goes unclaimed anyway
If the math is this favorable, why do most credits expire? Three reasons:
- Nobody is watching. The provider will not proactively credit you. You have to notice the breach, and the claim windows are short: roughly 60 days for AWS and Azure, and 30 days for Google Cloud.
- The math feels intimidating. Teams assume computing eligible uptime is hard. It is arithmetic: total minutes in the month, minus impacted minutes, over total minutes.
- The evidence ages out. Status dashboards and monitoring graphs roll off. If you wait a month to gather proof, the proof is gone. Our evidence checklist covers exactly what to capture and when.
How to turn this into collected dollars
The workflow is short. Confirm your measured uptime dropped below the SLA target. Identify which tier it lands in. Multiply that percentage against the affected service's monthly charge. File within the window with the exact figures. Our how-it-works overview lays out the end-to-end path, and our step-by-step claim guide covers the submission details for each provider.
If tracking outages by hand sounds tedious, that is the point of automation. Cloud Credits focuses on the billing-accuracy side of provider money, and you can see live outage data at Cloud Downtime. For fully automated detection, evidence capture, and claim drafting, Next Signal (our sponsor) watches your accounts and produces the claim for you.
The bottom line
SLA credits are worth claiming because the payout scales with your bill and the effort does not. A 10% credit on a five-figure service line is thousands of dollars for an hour of work, and severe breaches pay far more. The only way to lose is to let the window close. Run your numbers and see what one outage is actually worth to you.
