Reading a Hosting SLA: What Service Credits Actually Cover

“99.9% uptime guarantee” sounds like a promise. In practice it is a contract with a definition, a measurement method, an exclusion list, and a compensation formula — and the compensation is almost always service credit, not cash. This article walks through how to read a hosting SLA in ten minutes and calculate what an outage actually pays you, so you can tell a meaningful guarantee from a decorative one.

Uptime Percentages in Real Downtime

Advertised uptime	Downtime per month	Downtime per year
99.0%	7h 18m	3.65 days
99.5%	3h 39m	1.83 days
99.9%	43m 12s	8h 46m
99.95%	21m 36s	4h 23m
99.99%	4m 19s	52m 35s

Note how narrow the spread is between 99.9% and 99.99% on a monthly basis — 43 minutes versus 4. Only a handful of outages per year separate the two, which is why the credit payout matters more than the headline number.

The Credit Formula Most Contracts Use

Standard credit schedules are tiered on monthly uptime percentage:

• 99.9% to 100% — no credit
• 99.0% to 99.9% — 10% of monthly fee
• 95.0% to 99.0% — 25% of monthly fee
• Below 95.0% — 50–100% of monthly fee

Run the numbers. A $10/month VPS with four hours of downtime in a month sits at roughly 99.4% — which lands in the 10% tier. Your compensation is $1.00. For a single-site hobby project that is fine. For anything revenue-generating, it is negligible compared to the cost of the outage.

Exclusions That Void Your Claim

• Scheduled maintenance — usually excluded, and notice periods can be as short as a few hours.
• “Emergency” maintenance — broadly defined, frequently unannounced.
• Network outside the provider’s control — upstream transit issues are typically excluded.
• Client-side or application faults — including your own misconfiguration.
• DNS and domain issues — rarely covered even when caused by the provider’s nameservers.
• Force majeure — standard, and reasonable.

The exclusion list is where most claims die. Check it before you check the uptime number.

How Downtime Is Actually Measured

• Monthly vs annual window. Annual measurement lets a provider absorb a full bad month and still report compliance.
• Single probe vs multi-region. One monitoring location is easily gamed or misreported.
• ICMP ping vs service check. A host can respond to ping while your web service is down — ping-based SLA definitions are nearly worthless.
• Provider-side monitoring. If the only measurement is the provider’s own dashboard, you have no independent evidence.

Before purchasing, run an external monitor. A free service checking HTTP every minute gives you a log that beats any dispute.

Claim Process: What to Expect

1. Document the outage: start time, end time, and independent probe evidence.
2. Submit the credit request within the stated window — commonly 7 to 30 days after the incident. Miss this and you forfeit.
3. Expect credits to be applied to a future invoice, not refunded.
4. Expect the provider to investigate first; many claims are resolved only after a support escalation.
5. Keep the ticket reference and the credit confirmation for your records.

What Credits Do Not Cover

Service credits compensate the hosting fee, never the business impact. Lost sales, SEO ranking dips from crawl errors, and customer churn are all outside scope. If your risk exposure exceeds a few hundred dollars per hour, an SLA credit is not your risk mitigation — redundancy is.

That is why the practical complement to any SLA is a second instance in a different region or on a different provider. The arithmetic is covered in when to add a second VPS instead of upgrading, and provider-level reliability differences are in the comparison table.

A Ten-Minute SLA Checklist

• Is the uptime target measured monthly or annually?
• Is measurement internal, external, or self-reported?
• Does the definition cover service availability or just network reachability?
• What is the credit percentage at 99.0%, 95.0%, and below?
• How long is the claim window, and how must evidence be submitted?
• Are credits capped at the monthly fee?
• Which exclusions are broad enough to swallow most incidents?

Choosing Providers on SLA Substance

Providers that state a monthly measurement window, use external monitoring, and publish a clear credit schedule are the ones worth trusting. InterServer publishes standard service commitments and applies credits to the account without a lengthy dispute process, which keeps expectations simple for budget buyers.

View InterServer VPS plans and SLA terms →

Bottom Line

An SLA is a credit schedule, not an insurance policy. Read the measurement window, the exclusions, and the tier table; then compute the payout for a realistic outage on your plan. If the number is small — which it usually is on a $5–$10 VPS — treat it as a minor goodwill gesture and invest in redundancy instead.