The Hidden Refund: How to Claim SLA Credits from AWS, Azure, GCP, and DigitalOcean
July 20, 2026
Your cloud provider promises 99.9% to 99.99% uptime in a legally binding Service Level Agreement. When they miss that number, they owe you service credits. There is a catch: nobody at AWS, Azure, Google Cloud, or DigitalOcean is going to send you that money unprompted. Credits are claim-only. If you don't file, the breach disappears into the same void as the outage.
Industry estimates consistently put unclaimed SLA credits in the billions of dollars per year. This guide walks through exactly how to claim yours: thresholds, deadlines, evidence, and the submission channel for each of the big four providers.
First, three rules that apply everywhere
- Credits are not cash. Every provider pays out as service credits applied to a future bill. Nobody mails a check.
- You must file the claim yourself. Providers detect their own outages just fine. They simply don't volunteer compensation. Miss the claim window and the credit is gone for good.
- Credits are calculated on the affected service, not your whole bill. A 10% credit on your EC2 bill means 10% of what you spent on EC2 instances in the breached region that month, not 10% of your total AWS invoice.
AWS: Support case within 60 days
AWS publishes per-service SLAs rather than one blanket agreement. The ones that matter most:
| Service | Uptime commitment | Credit tiers |
|---|---|---|
| EC2 (multi-AZ) | 99.99% | 10% below 99.99% · 30% below 99.0% · 100% below 95.0% |
| S3 | 99.9% | 10% below 99.9% · 25% below 99.0% · 100% below 95.0% |
| RDS (multi-AZ) | 99.95% | 10% below 99.95% · 25% below 99.0% · 100% below 95.0% |
| Lambda | 99.95% | 10% below 99.95% · 25% below 99.0% · 100% below 95.0% |
How to file: Open an AWS Support case with "SLA Credit Request" in the subject. Include your account ID, the billing cycle, the region, the exact dates and times (with time zone) of the impacted period, and request logs that document the errors. AWS explicitly asks for evidence; redact anything sensitive. Individual credits under $1 are not paid out.
Deadline: Within 60 days of the end of the billing cycle in which the incident happened. In practice: file in the month after the outage, not "when you get around to it."
We wrote a dedicated walkthrough with a copy-paste letter: AWS SLA Credits: Exact Thresholds, Deadlines, and a Copy-Paste Claim Template.
Microsoft Azure: Portal support request within 60 days
Azure SLAs are also per-service:
| Service | Uptime commitment | Credit tiers |
|---|---|---|
| Virtual Machines (Availability Zones) | 99.99% | 10% below 99.99% · 25% below 99.0% · 100% below 95.0% |
| App Service | 99.95% | 10% below 99.95% · 25% below 99.0% · 100% below 95.0% |
| SQL Database | 99.99% | 10% below 99.99% · 25% below 99.0% · 100% below 95.0% |
| Cosmos DB | 99.999% | 10% below 99.999% · 25% below 99.99% · 100% below 99.9% |
How to file: In the Azure portal, open Help + support → New support request. Set the issue type to Billing and the problem type to Refund Request. Azure support will ask for your tenant GUID, the outage incident ID (from the Azure status page or Service Health), proof of impact, the number of affected users or workloads, and what you tried to resolve it.
Deadline: Within 60 days of the end of the month in which the incident occurred.
The one rule that decides whether you ever see a credit: claims are filed by you, and the clocks are short. AWS and Azure give you 60 days. Google Cloud and DigitalOcean give you 30. Miss the window and the credit is gone for good.
Google Cloud: SLA contact form within 30 days
GCP calls them Financial Credits, and its claim window is the shortest of the four: 30 days.
| Service | Uptime commitment | Credit tiers |
|---|---|---|
| Compute Engine (multi-zone) | 99.99% | 10% below 99.99% · 25% below 99.0% · 50% below 95.0% |
| Cloud Storage | 99.95% | 10% below 99.95% · 25% below 99.0% · 50% below 95.0% |
| Cloud SQL | 99.95% | 10% below 99.95% · 25% below 99.0% · 50% below 95.0% |
| BigQuery | 99.99% | 10% below 99.99% · 25% below 99.0% · 50% below 95.0% |
How to file: Use Google's Cloud Platform SLA contact form (search "GCP SLA credit" or go through Cloud Support). Submit once per product, per month: batching multiple incidents for the same product into one claim is expected. Google requires the request to come from a Google or Google Workspace account email.
Deadline: Within 30 days of the end of the month in which the incident occurred. This is the window most teams miss.
DigitalOcean: Support ticket within 30 days
DigitalOcean keeps it simpler than the big three:
| Service | Uptime commitment | Credit tiers |
|---|---|---|
| Droplets | 99.99% | 5% below 99.99% · 10% below 99.0% · 30% below 95.0% |
| Spaces | 99.99% | 5% below 99.99% · 10% below 99.0% · 30% below 95.0% |
| Managed Databases | 99.99% | 5% below 99.99% · 10% below 99.0% · 30% below 95.0% |
How to file: Open a support ticket from your DigitalOcean account. Name the affected resources (Droplet IDs, database cluster names), the region, and the impact window with timestamps.
Deadline: Within 30 days of the end of the month in which the incident occurred.
The evidence checklist that makes claims stick
Regardless of provider, approved claims tend to include the same five artifacts:
- Timestamps of the outage window in a stated time zone (UTC is safest).
- The vendor's own incident record: a link to the status-page incident or incident ID.
- Your affected resource identifiers: account ID, subscription/tenant ID, project ID, region, instance or cluster IDs.
- Impact proof: error logs, failed-request counts, screenshots of 5xx responses.
- The SLA clause you're claiming under, cited by name (e.g. "Amazon EC2 Service Level Agreement, Monthly Uptime Percentage").
Missing the incident ID or the resource identifiers is the most common reason claims stall in support limbo.
Why almost nobody collects
The workflow above is maybe an hour of work per incident, if you noticed the outage, matched it to your services, looked up the right SLA, and remembered the deadline. Multiply that by every service and region you run, across four providers, every month. That's the real cost: not the claiming, but the tracking.
That's the gap UptimeAudit closes. We monitor the big four providers' status pages and health endpoints down to the service and region level; see the live status dashboard, and when an observed outage crosses your provider's SLA threshold, we draft the claim for you: pre-filled with your account details, the incident data, and the deadline. You review it and submit it through your provider's normal support channel. Nothing is sent without your approval, and every claim is still decided by the provider under its SLA terms.
The refunds are real. They just don't file themselves.