Hosting Uptime, SLAs and Support - What the Guarantees Really Mean
How to read hosting uptime percentages and service level agreements, what downtime they allow, how credits work, and how to judge support quality.
An uptime guarantee of 99.9 percent allows about 43 minutes of downtime in a 30-day month, 99.99 percent allows about four minutes, and the remedy for missing it is usually a small account credit rather than compensation for lost business. The number matters less than how downtime is measured, what is excluded and how quickly support helps. This guide shows how to translate percentages into minutes, read an SLA and test support before you depend on it.
Uptime percentages in minutes
Uptime is the share of time a service is available over a period. The arithmetic is simple: allowed downtime = (1 - uptime) x total time.
For a 30-day month there are 30 x 24 x 60 = 43,200 minutes.
| Uptime promise | Allowed downtime per 30-day month | Per year (365 days) |
|---|---|---|
| 99 percent | About 432 minutes (7.2 hours) | About 3.65 days |
| 99.5 percent | About 216 minutes (3.6 hours) | About 1.8 days |
| 99.9 percent | About 43 minutes | About 8.8 hours |
| 99.95 percent | About 22 minutes | About 4.4 hours |
| 99.99 percent | About 4.3 minutes | About 53 minutes |
A difference between 99.9 and 99.99 percent is the difference between roughly 8.8 hours and under an hour of downtime in a year, which matters for a store and may be irrelevant for a hobby blog.
What a service level agreement covers
A service level agreement, or SLA, is the part of the contract that states the uptime target, how it is measured and what happens if the provider misses it. Read these points:
- Scope. Does the SLA cover only the network and power, or also the server and the web service? Some cover only the infrastructure.
- Measurement period. Monthly is common. A longer period hides short bad stretches.
- Exclusions. Scheduled maintenance, events outside the provider's control, issues caused by your software or abuse, and DDoS attacks are frequently excluded.
- Claim process. Many providers require you to request the credit within a limited window, and some require proof.
- Remedy. Usually service credit, often a percentage of the monthly fee tied to how far uptime fell below the target. Credits are rarely worth more than your monthly fee and never cover your lost sales.
- Plan coverage. Some SLAs apply only to certain plans. Entry shared plans may carry a softer commitment than business-tier or cloud plans.
Worked example: valuing downtime
Illustrative. An online shop earns $60,000 a year, about $164 a day on average, which is about $6.85 an hour averaged across the day. Sales are not evenly distributed, so assume peak hours are three times the average, or about $20.50 an hour.
- At a 99.9 percent guarantee, allowed downtime is about 8.8 hours a year. If it all happened at peak time, lost sales could be about 8.8 x $20.50 = $180.
- At 99.5 percent, allowed downtime is about 1.8 days, or about 43 hours. In the same peak-time scenario, lost sales could be about 43 x $20.50 = $882.
The difference of about $700 a year helps you judge whether a plan with a higher uptime commitment is worth an extra $5 to $10 a month. For a shop of this size, it probably is; for a site with a few hundred visits a month, it probably is not. These numbers rest on assumptions, so use your own.
Judging reliability beyond the promise
- Track actual uptime. Use an independent uptime monitor so you have your own data.
- Look at the provider's status page. Does it exist, is it updated promptly and does it explain incidents?
- Read independent reviews with care. Look for patterns across many reports, and remember that reviews can be affiliate-driven.
- Check redundancy. Ask about backup power, network providers and data center standards, and whether the service has redundant components.
Evaluating support
Support quality matters most when something is wrong. Before you commit:
- Check the channels. Chat, tickets and phone. Which are available on your plan and at what hours?
- Ask a real technical question during the trial or refund window, and note how long it takes to get a useful answer.
- Read the knowledge base. A thorough, current knowledge base reduces how often you need support.
- Understand the scope. Support for your own code, plugins or third-party issues may be limited. Ask where the boundary is.
- Find the escalation route. Is there a way to reach a senior technician if first-line support fails?
| Support signal | Good sign | Warning sign |
|---|---|---|
| Response time | Minutes on chat or hours on tickets | Days to first reply |
| Answers | Specific, with steps | Scripted, repeating the question |
| Ownership | Follows up until resolved | Closes tickets quickly |
| Documentation | Current and searchable | Outdated screenshots |
| Incident communication | Status page and email | Silence |
Mistakes to avoid
- Assuming all downtime is covered by the SLA.
- Forgetting to claim credits within the window.
- Paying for a higher-priced plan for uptime without checking the SLA applies to it.
- Testing support only after an emergency.
- Having no independent monitoring, so you learn about outages from customers.
Frequently asked questions
Is 99.9 percent uptime good enough?
For most small business sites it is a reasonable standard. A store with high revenue per hour may want a higher commitment plus its own resilience measures.
Do I get money back when my host is down?
Usually only a service credit toward future bills, and often only if you claim it. It rarely reflects your actual losses.
How do I monitor uptime?
Use a monitoring service that checks your site from outside your host and alerts you by email or text. Many offer free tiers.
What is the difference between uptime and speed?
Uptime says the site is available. Speed describes how quickly it responds. A site can be up and unusably slow.
Sources and further reading
- Federal Trade Commission - advertising and service claims.
- CISA - resilience and incident planning for small organizations.
- web.dev - performance measurement guidance.
