SaaS Outage Credits: Build a Claim That Matches Your Agreement
By Anirudh Menon, Viaclause editor
Connect outage evidence, eligible charges and filing requirements in one record, then track any approved service credit through to the invoice.
AI-generated editorial illustration by Viaclause. The person and setting are fictional.
After a software outage, open a claim record before you calculate a credit. Identify the agreement, the affected service and the submission route. Then collect evidence that your account was affected. A public incident announcement alone does not tell you what your business can recover.
Use the vendor breach response record to connect the facts, governing clause and proposed response.
For a procurement team, the useful question is: can we connect the incident to an eligible charge and a complete, timely request? The following workflow is our proposed way to make that connection. It is not a finding that a vendor owes you money.
Separate the outage from the credit entitlement
The Amazon Compute SLA distinguishes regional and individual-instance commitments. It requires a support claim with incident details, resource identifiers and supporting logs. Its deadline is the end of the second billing cycle after the incident, not a universal 60-day period. Credits depend on the applicable calculation and exclusions. EC2 is a cloud infrastructure example; its rules do not become the rules of a SaaS product running on it.
Check the SaaS liability and indemnity claim map if the credit provision interacts with other remedies or caps.
The Microsoft Partner Center credit instructions illustrate another distinction. In that partner purchasing context, the affected service within a suite matters. Microsoft asks for evidence of customer impact and a request from the customer, not just a ticket identifier. Check whether you bought directly or through a partner before choosing a submission route.
First establish who sold you the service. If your software supplier uses AWS, that does not by itself make your company the AWS customer entitled to submit the infrastructure claim. Start with your own supplier agreement and account records.
Keep evidence, terms and completion together
Copy these fields into your incident file. Keep technical evidence and commercial instructions together, with a named owner for each missing item.
- Technical record: affected account and service; resource or tenant identifiers; start and end timestamps with time zone; failed operation; monitoring or error-log location; provider incident identifier; what remained available; recovery steps.
- Commercial record: seller and purchasing route; signed order and applicable SLA version; eligibility conditions; measurement period; excluded downtime; eligible charge; claim deadline and its starting event; submission channel; required attachments; person authorized to file.
- Completion record: submission timestamp; case number; acknowledgment; follow-up request; decision; credit amount; invoice on which it appeared; unresolved difference.
Write “unknown” where evidence is missing. For example, “account affected, confirmed by internal logs; contractual outage calculation, awaiting IT review” is more useful than treating every reported incident minute as eligible downtime.
A worked example: the affected charge changes the answer
This is a hypothetical budgeting example, not a vendor entitlement calculation. Assume an agreement awards 10% of the eligible monthly service charge after all conditions are met. Your invoice totals $4,000, but only $1,200 belongs to the eligible affected service.
Provisional credit: $1,200 × 10% = $120. Applying 10% to the whole invoice would produce $400 and overstate the request by $280. Neither figure proves eligibility. Taxes, one-time charges, exclusions and contractual caps still need their own checks.
For a different agreement, replace both inputs. Do not borrow a provider's percentage while using another provider's charge definition. Where several incidents affect the same service in a month, first check whether the agreement uses a monthly aggregate, an incident calculation or another method. Avoid counting the same charge twice.
Preserve the deadline without sending an incomplete claim
Record the actual deadline wording and translate it into a date with the contract owner. If a calculation is uncertain, request clarification early through the specified route. Do not assume that opening an ordinary support ticket preserves a separate credit claim.
Use this internal handoff message:
Incident [reference] affected [account/service] during [times and zone]. Evidence is saved at [approved location]. The applicable SLA is [version/link], and the claim requirement is [clause]. Please confirm the eligible charge, calculation and submission deadline by [internal date]. [Name] owns filing; [name] will match any approved credit to the invoice. Unresolved items: [list].
Before attaching logs, remove secrets and unnecessary personal or customer data. AWS's claim instructions specifically call for masking sensitive information. Preserve the original evidence securely; send the appropriately redacted copy.
Finish at the invoice, not the acknowledgment
Track “filed,” “approved” and “applied” separately. A hypothetical $120 request is not $120 recovered until the adjustment is confirmed and reconciled. If the vendor denies the request, ask which condition or exclusion was not met and preserve its explanation.
Keep a separate record of the business disruption. AWS's remedy language makes the SLA the exclusive remedy unless the governing agreement provides otherwise. A credit calculation should not silently become a claim for all business losses. For a wider dispute, prepare a focused adviser packet. If repeated incidents affect your renewal decision, assign that decision through your renewal ownership process.
Method: We inspected the vendor documents below. The workflow and arithmetic example are Viaclause's proposed aids; we did not file or test a claim.
Sources
- Amazon Compute Service Level Agreement, AWS, updated May 25, 2022; checked September 27, 2026. EC2 scope, claim requirements, calculation, exclusions and remedy limits.
- Request a credit from Microsoft, Microsoft Learn, updated June 16, 2026; checked September 27, 2026. Partner Center purchasing context, affected-service scope and evidence requirements, not a universal direct-customer SLA.