Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

SLA service credits: what they compensate, what they do not, and how to claim one

Technical article
Hosting

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

A service credit is a reduction in a future bill. In the published agreements we read, it is a percentage of one month's fee for the affected service, it is paid only after you claim, and it is stated to be the customer's sole remedy under the SLA for the outage (the wider agreement and applicable law may still matter). It is a price adjustment for a service that fell short. It is not insurance, and it is not sized to what the outage cost your business.

So the practical comparison is simple. Put the credit you would receive next to what one day of outage costs you, and treat the gap as risk you hold unless you reduce it with design, contracts or insurance. This guide shows how that gap arises, using public agreements as examples. It does not cover how availability percentages are measured; see the companion article on reading an availability figure in a service report.

This is general information, not legal advice.

Why it matters

Buyers often read '99.99 per cent SLA' as a promise of near-perfect uptime with a penalty for the provider if it fails. Microsoft's guide to reading agreements corrects that: "SLAs don't guarantee that your application is reliable, that your users avoid downtime, or that the service provider prevents all failures" (Microsoft, How to read a service-level agreement). The guide describes the SLA as a set of conditional commitments, with compensation "usually in the form of service credits".

What the credit pays is therefore a share of the fee. If the affected service costs a few hundred dollars a month and your business loses a day of sales, the credit and the loss are not the same order of size (our illustration, with an invented example below).

How the credit is calculated

The agreements we read set out tiers. Each tier ties a monthly uptime percentage to a credit expressed as a percentage of a monthly charge.

Agreement (as examples) Credit tiers, by monthly uptime
Amazon EC2, region level below 99.99% and at least 99.0%: 10%; below 99.0% and at least 95.0%: 30%; below 95.0%: 100%
Amazon EC2, single instance below 99.5% and at least 99.0%: 10%; below 99.0% and at least 95.0%: 30%; below 95.0%: 100%
Amazon S3, standard classes below 99.9% and at least 99.0%: 10%; below 99.0% and at least 95.0%: 25%; below 95.0%: 100%
Google Compute Engine, instances in multiple zones (Premium Tier, most regions) 99.00% to below 99.99%: 10%; 95.00% to below 99.00%: 25%; below 95.00%: 100%
Azure Virtual Machines across availability zones (Microsoft agreement, October 2026 edition) below 99.99%: 10%; below 99%: 25%; below 95%: 100%

Sources: AWS Compute SLA, AWS S3 SLA, Google Compute Engine SLA, and the Microsoft Service Level Agreement for Microsoft Online Services. These are examples of the pattern. Other tiers and services in each agreement differ, and a hosting provider's own agreement may be quite different.

What the percentages mean in time

Convert the thresholds to minutes. In a 30-day month there are 43,200 minutes. Using our arithmetic:

  • 99.99 per cent allows about 4.3 minutes of downtime.
  • 99.0 per cent allows about 7.2 hours.
  • 95.0 per cent allows 36 hours.

So in the EC2 region-level agreement, the first credit tier of 10 per cent covers a month with anything from about 4.3 minutes up to about 7.2 hours of downtime. A full day of outage in that month sits in the 30 per cent tier. A full credit of 100 per cent begins only after more than 36 hours of unavailability. These are our conversions of published thresholds, and they depend on how each agreement defines downtime.

What the credit is a percentage of

The base of the percentage matters as much as the percentage.

  • AWS says credits are "calculated as a percentage of the monthly bill" for the affected service in the affected region, or for the single instance that missed its commitment (Compute SLA). For S3, the base is the charges for the affected storage class in the affected region.
  • Google applies the percentage to the "monthly bill for the respective Covered Service in the Region that did not meet SLO".
  • Microsoft says credits "apply only to fees paid for the particular Service, Service Resource, or Service tier for which a Service Level has not been met", "only with respect to the actual downtime and not the overall outage duration", and that credits in a billing month "will not, under any circumstance, exceed your monthly service fees" for that service or resource.

In each case the credit is bounded by the fee for the affected service. It is not linked to the revenue that depended on it, the cost of staff time, or the loss caused by an interrupted application running on top. That is the main reason a credit differs from compensation for loss.

The form of the credit

The agreements describe credits as bill reductions, not payments:

  • AWS: "We will apply any Service Credits only against future payments for Amazon EC2 otherwise due from you", and "Service Credits will not entitle you to any refund or other payment from AWS." It also says credits are issued only if the amount for the month is "greater than one dollar ($1 USD)".
  • Google: credits are "applied to future use of the Covered Service" and the maximum in a month cannot exceed what the customer owes for the affected services.
  • Microsoft: credits are applied to the monthly service fees, and the customer "may not unilaterally offset" its Applicable Service Fees for performance or availability issues. A customer who withholds payment because of an outage is acting outside the agreement's stated process (our reading).

If you leave the service, a credit against future bills may be of no value to you (our reading of "future payments").

Claims: you must ask, in time, with evidence

Credits are not paid automatically. Microsoft's guide says providers "don't usually apply them automatically" and that you "must detect outages that breach the SLA, document them, and submit claims within the deadline" (Microsoft guide, above; the second phrase appears in the guide's section on claiming credits).

The deadlines in the three agreements differ:

  • AWS: the claim "must be received by us by the end of the second billing cycle after which the incident occurred", and must include logs and details of each incident. Failure to provide them "will disqualify you from receiving a Service Credit".
  • Google: the customer must notify support "within 60 days from the time Customer becomes eligible to receive a Financial Credit" and provide log files showing downtime periods. If not, "Customer will forfeit its right to receive a Financial Credit".
  • Microsoft (agreement, October 2026 edition): for Azure, "we must receive the claim within 60 days of the Incident". The claim must include a description of the incident, the time and duration of downtime, affected resource names, affected users and a description of errors, and "Failure to provide the required information will result in claims being rejected."

Two practical consequences follow. First, you need your own records. If the only evidence of downtime is the provider's own monitoring, you depend on the provider's account of the month, so keep your own timestamped log of outages (our recommendation). Second, the deadline starts at the incident or the end of the billing period, not when you notice, so someone should own the monthly check.

Sole remedy and exclusions

All three agreements contain wording that makes the credit the exclusive remedy. AWS's EC2 text says "this SLA sets forth your sole and exclusive remedies" unless the wider agreement provides otherwise. Google says the SLA "states Customer's sole and exclusive remedy for any failure by Google to meet the SLO". Microsoft's text says "Service Credits are your sole and exclusive remedy for any performance or availability issues for any Service under the Agreement and this SLA", and adds: "Service Credits will not be awarded to compensate for any other forms of loss, including but not limited to lost revenue, operational costs, or any indirect losses experienced by you or the end-users."

Taken at face value, that means the SLA channel does not pay for lost sales or recovery costs. Whether other claims are available depends on the wider contract and on law, which is a matter for a lawyer.

Credits also depend on the outage counting at all. Exclusions such as events outside the provider's reasonable control, customer actions, preview features and planned maintenance can remove an outage from the calculation entirely. Those are covered in the companion article.

If you buy through a reseller or hosting partner, the claim route may differ from the original vendor's. Ask who receives the claim and who pays the credit (our recommendation).

Contract protections in Australia

For a small business buying under a standard form contract, Australian law has rules about unfair terms. The ACCC says "Laws protect consumers and small businesses from unfair terms in standard form contracts" and that, from 9 November 2023, proposing, using or relying on unfair terms in such contracts is banned and penalties apply. A small business is covered if it has fewer than 100 employees or less than $10 million in annual turnover (ACCC, Contracts). The ACCC also says "Only courts can make final decisions about whether a contract term is unfair."

We do not say that any credit or sole-remedy term is or is not unfair. That depends on the whole contract and is for a court to decide. The point for a buyer is limited: if you sign a standard form contract with remedies you do not understand, seek advice before relying on them. This is general information, not legal advice.

What to choose and what to agree

This section is our recommendation.

  1. Treat the credit as a price adjustment, not protection. Plan continuity by design: backups, replication and failover are separate things, and none of them is purchased by a credit.
  2. Ask for the schedule in writing. The tiers, the base for the percentage, any cap and the form of payment.
  3. Agree who claims. Name the person who checks the monthly report and files claims before the deadline.
  4. Keep your own evidence. An independent check of your own application, with timestamps.
  5. Decide your risk. If a day's outage costs more than you can absorb, cover the gap with design, a different contract or insurance, and do not rely on the credit.

Questions to put to your provider

  1. What is the credit schedule, and what is each percentage calculated on?
  2. Is there a cap, and does it include all services or only the affected one?
  3. How is the credit given: bill reduction, cash or service time?
  4. How do I claim, what evidence is needed, and by when?
  5. Do you notify me when the month falls below the commitment, or must I find it?
  6. Is the credit the only remedy, and what does the wider contract say about liability?
  7. If I buy through a partner or reseller, who pays the credit and who do I claim from?

A worked example (illustrative)

The following is invented. A business pays $600 a month for a hosted server whose agreement gives a 10 per cent credit if monthly uptime falls below 99.99 per cent and a 25 per cent credit below 99.0 per cent. The server is down for a 24-hour outage, one full day. In a 30-day month that is roughly 3.3 per cent downtime, so uptime is about 96.7 per cent.

That sits below 99.0 per cent and above 95.0 per cent, so the 25 per cent tier applies: a credit of $150. The business estimates that a day without its order system cost it $9,000 in lost orders and staff time. The credit is under two per cent of the loss. The business claims on the 10th of the next month, with its own downtime log, because the claim deadline is 60 days.

The business then reviews whether a second server or a tested failover would cost less than the risk it holds.

Risks and limits

  • We read three published infrastructure agreements, the Microsoft one being a consolidated document, the October 2026 edition. Terms change, and your agreement may differ.
  • The percentages, tiers and deadlines above are examples. A hosting provider's agreement is its own; check yours for the actual percentages, tiers and deadlines.
  • Our time conversions assume a 30-day month and the definitions in each agreement.
  • The legal paragraph is a summary of regulator guidance, not a view on any clause.

Next step

Compare the credit in your agreement with the cost of one day of outage for your most important system, and write down the gap. Then confirm who owns the monthly check and the claim deadline. If you are reviewing a hosted service, the Cloud Hosting page is a starting point for that conversation; ask for the scope in writing.

Sources

All accessed 15 November 2026.

Learning notes

  • A service credit is a discount on a future bill, not a payment for your loss.
  • The percentage is of the affected service's fee, and it is capped by that fee.
  • You usually have to claim, with evidence, before a deadline.
  • 'Sole and exclusive remedy' wording makes the credit the SLA remedy for an availability failure, subject to the wider agreement and applicable law.
  • Compare the credit with one day of your own outage cost.