A 99.9% uptime guarantee allows about 43 minutes of downtime in a 30-day month, or almost 8 hours and 46 minutes a year. And if a host misses it, you usually get a small credit on your bill, not compensation for lost sales. The number is a useful starting point, but how the host measures it, and how fast they fix problems, matter more.
Key takeaways
- 99.9% uptime still allows nearly nine hours offline a year. 99.99% allows under an hour.
- An SLA’s remedy is usually a service credit, and you often have to claim it yourself.
- Exclusions such as scheduled maintenance and problems in your own website code can shrink the promise.
- Independent monitoring is the only way to know whether a host meets its number.
- Fast, local support matters more than an extra nine after the decimal point.
What does 99.9% uptime mean in hours?
Each extra nine cuts the allowed downtime by 90%. Here’s what common uptime figures allow:
| Uptime | Downtime in a 30-day month | Downtime in a year |
|---|---|---|
| 99% | 7 hours 12 minutes | 3 days 15 hours |
| 99.5% | 3 hours 36 minutes | 1 day 20 hours |
| 99.9% | 43 minutes | 8 hours 46 minutes |
| 99.95% | 22 minutes | 4 hours 23 minutes |
| 99.99% | 4 minutes | 53 minutes |
For most small business websites, 99.9% is a fair standard. The difference is when the downtime happens. Forty minutes at 3am on a Sunday is barely noticed. Forty minutes during a Saturday sale is lost revenue.
What does an uptime guarantee actually promise?
An uptime guarantee is part of a service level agreement (SLA). It’s a promise to credit part of your bill if the service falls below the target. It isn’t insurance for lost sales.
Amazon’s EC2 SLA is a clear example. Below the target, customers get a credit of 10%, 30% or 100% of that month’s charge, depending on how bad the outage was. The credit must be requested within a set time, and AWS says the SLA is the customer’s “sole and exclusive” remedy (AWS). Smaller hosts often work the same way, with less generous terms.
The fine print to look for
- Exclusions. Scheduled maintenance, attacks and problems caused by your own website are often left out.
- Measurement. Is it measured monthly or yearly? A yearly figure lets one long outage hide among good months.
- Claims. Do you have to apply, with evidence, before a deadline?
- The cap. Most credits are capped at the fee for that month.
What counts as downtime?
This is where owners and hosts often disagree. The host measures whether its server is responding. You care whether customers can see your pages and buy.
A server can be “up” while your WordPress site shows a critical error, a database connection error or a blank page. To the host, that’s your website’s problem, not downtime. That’s why website support for the site itself matters as much as the server.
How do I know if my host keeps its promise?
Measure it yourself. Independent uptime monitoring checks your site from outside and records every outage. Check how often it looks, though. UptimeRobot’s free plan checks every five minutes, while its paid plans go down to 15 seconds (UptimeRobot). A five-minute check can miss a short outage completely.
Our guide to website uptime monitoring explains what a good monitor should check and how to set up alerts. Keep the records. If you ever need to claim a credit or change hosts, they’re your evidence.
What matters more than the uptime number?
How quickly someone fixes the problem. A host promising 99.99% that takes a day to answer a ticket will hurt you more than a 99.9% host who picks up the phone.
Long outages also affect search. Google says a site should only be fully unavailable for “a few days at most”, and recommends a 503 status for short planned downtime (Google Search Central). It warns that being removed from the index “can take quite some time to recover from”.
Ask about support hours and who answers. Our guides to web hosting support and Australian based hosting support cover the red flags and why time zones matter.
How should I compare uptime guarantees?
- Convert the percentage to hours. Use the table above, so you know what you’re actually being promised.
- Read the exclusions. Check what doesn’t count, especially maintenance windows.
- Ask how they measure it. Find out whether they check the server, the network or your actual website.
- Check the remedy. Find out how credits work, and whether you must claim them.
- Test support before you buy. Send a pre-sales question and see how fast a real person answers.
What do we commit to?
Our hosting legal pages include a 99.9% uptime SLA and a 30-day money-back guarantee. Sites run in NEXTDC data centres in Brisbane and Melbourne. Support depends on the plan: email, ticket and chat on Personal; phone support 7 days, 9 to 5, on Business; and 8am to 8pm, 7 days, with a 4-hour response on VPS. Our 24/7 emergency line is 0426 833 422.
You can compare our hosting plans, or read more about our Australian hosting support and who you’ll speak to.
Frequently asked questions
How much downtime is 99.9% uptime?
About 43 minutes in a 30-day month, or about 8 hours and 46 minutes over a year. It sounds small, but it can all happen at once, so check how the host measures it and how quickly they respond.
What happens if my host doesn’t meet its uptime guarantee?
Usually you can claim a credit on a future bill, often a percentage of that month’s fee. Most SLAs say the credit is your only remedy, so it won’t cover lost sales. Many require you to claim within a set time and supply evidence.
Is 100% uptime possible?
Not in practice. Hardware fails, networks have problems and software needs updates. A host that promises 100% usually means it will credit you when it misses, not that downtime never happens. Treat it as a marketing figure and read the SLA.
Does scheduled maintenance count as downtime?
Often not. Many SLAs exclude planned maintenance, provided the host gives notice. Ask how much notice they give, how often maintenance happens and whether it’s scheduled outside Australian business hours.
Do I need 99.99% uptime for a small business site?
Rarely. For most brochure sites and small shops, 99.9% with fast local support is enough. Higher figures usually need redundant servers and cost much more. Very busy online stores are the main exception.
Sources
- Amazon Compute Service Level Agreement, AWS, updated 25 May 2022
- Pricing, UptimeRobot, viewed November 2026
- Pause or disable a website, Google Search Central