Hosts advertise uptime as a percentage. The numbers look almost the same, but the downtime they allow is very different.
| Uptime | Downtime per month | Downtime per year |
|---|---|---|
| 99% | about 7 hours 18 minutes | about 3 days 15 hours |
| 99.5% | about 3 hours 39 minutes | about 1 day 20 hours |
| 99.9% | about 44 minutes | about 8 hours 46 minutes |
| 99.95% | about 22 minutes | about 4 hours 23 minutes |
| 99.99% | about 4 minutes 23 seconds | about 52 minutes |
Months here are the average month, a twelfth of a year.
How the numbers are worked out
The arithmetic is simple. A year has about 8,766 hours (365.25 days). Multiply by the share that is allowed to be down:
8,766 hours × (100% − 99.9%) = 8,766 × 0.001 ≈ 8.8 hours a year
Each extra 9 cuts allowed downtime by ten. That is why 99.99% (“four nines”) is a far bigger promise than 99.9%: about 53 minutes a year instead of nearly 9 hours.
What an uptime guarantee usually promises
An uptime guarantee, part of a service level agreement (SLA), is a promise of compensation, not of uptime. If the host misses the target, you typically get a credit off next month’s bill, often a small percentage, and usually only if you claim it.
Read the details:
- What counts as down? Often only the host’s network or hardware. Your site crashing because of its own software does not count.
- Planned maintenance is usually excluded.
- How is it measured? The host’s own monitoring, from their own network, is common.
Your site’s uptime is not your host’s uptime
A host can hit 99.99% on its network while your site is down far more often. Most website outages are not hosting failures at all:
- An expired SSL certificate or domain.
- A plugin or theme update that causes a fatal error.
- A full disk, or the database killed for lack of memory.
- A DNS change gone wrong.
- A traffic spike the server cannot handle.
None of these counts against the host’s SLA. Your real uptime depends as much on how the site is run as where it is hosted.
Getting more nines
Each step up costs more and adds complexity. Roughly:
| Target | What it takes |
|---|---|
| 99% to 99.9% | A reputable host, monitoring, automatic certificate and domain renewal, tested updates |
| 99.9% to 99.95% | Page caching and a CDN to absorb spikes, fast restore from backups |
| 99.95% to 99.99% | More than one server behind a load balancer, a replicated database, no single point of failure |
For most sites, the first row matters most. Getting there is mostly about avoiding self-inflicted outages.
Why it still matters
The figure tells you how a host designs its systems. Reaching 99.99% needs redundant power, network paths and hardware. A host promising 99% is telling you outages of several hours are possible.
Measuring your own uptime
Use an external monitor that checks your site every minute or so from outside your host’s network and alerts you when it fails. Many free services do this. The record they keep is far more useful than a percentage on a sales page.
Claiming SLA credits
If your host misses its target, credits are rarely automatic. Note the start and end times of each outage, keep your monitoring reports as evidence, and submit a claim within the window the SLA sets, often 30 days. Expect a partial credit on the hosting fee, not compensation for lost sales.
Related
Something out of date? Software changes. If a step no longer works, tell us and we will check it and update the page.

