Every hosting plan promises it. “99.9% uptime guaranteed.” It sounds close to perfect, and the difference between 99.9% and 99.99% looks like a rounding error. It isn’t. That one extra nine is the difference between about nine hours of downtime a year and under one.
This guide explains what uptime percentages actually mean, how much downtime each one allows, what’s hiding in hosting guarantees and how to measure your real uptime instead of trusting the brochure.
What uptime means
Uptime is the share of time a service is available and working, expressed as a percentage:
Uptime % = (total time − downtime) ÷ total time × 100
If your site was unreachable for 43 minutes in a 30-day month, that’s 43 out of 43,200 minutes, or 99.9% uptime.
Engineers often talk in “nines”: two nines is 99%, three nines is 99.9%, four nines is 99.99%, and so on.
How much downtime each level allows
| Uptime | Downtime per year | Per 30-day month | Per week |
| 99% | 3 days 15 h 36 min | 7 h 12 min | 1 h 41 min |
| 99.5% | 1 day 19 h 48 min | 3 h 36 min | 50 min |
| 99.9% | 8 h 46 min | 43 min | 10 min |
| 99.95% | 4 h 23 min | 22 min | 5 min |
| 99.99% | 53 min | 4 min 19 s | 1 min |
| 99.999% | 5 min 15 s | 26 s | 6 s |
Each extra nine cuts allowed downtime by a factor of ten, and roughly multiplies the cost and effort of achieving it. 99.9% is achievable on a single good server with sensible practices. 99.99% usually needs redundancy across servers, automatic failover and someone who responds to alerts within minutes, day or night. 99.999% is the territory of telecoms and payment networks.
Which target do you actually need?
- Personal site, portfolio or blog: 99.5–99.9% is fine. Nobody loses money if it’s down for an hour on a Sunday.
- Small business site: 99.9%. Downtime costs enquiries, but a short outage overnight is survivable.
- Online shop: 99.95% or better, especially in peak seasons. Every minute down at checkout is lost revenue.
- SaaS product or API: 99.9–99.99%, depending on what your customers build on top of you. Your uptime becomes part of theirs.
Pick a target you can actually meet. A 99.99% promise you miss every quarter is worse than an honest 99.9% you hit.
Why your real uptime is lower than your host’s
Uptime multiplies. If your site depends on a web server, a database and a third-party payment API, each at 99.9%, the chance they’re all up at once is roughly 99.9% × 99.9% × 99.9% ≈ 99.7%. That’s about 26 hours of potential downtime a year, three times what any single component promises.
The host’s guarantee covers the host. Your plugin update that broke the site, your expired SSL certificate or your lapsed domain don’t count against their 99.9%. To your visitors, they’re all downtime.
“Guaranteed” means credits, not uptime. Most SLAs refund a percentage of your monthly fee when they miss the target. A $10 refund doesn’t cover a day of lost sales. Read the fine print, too: planned maintenance is usually excluded, and you often have to claim the credit yourself within a deadline.
How uptime is measured (and how it can mislead)
Uptime figures are only as good as the measurement behind them.
- Check interval. A monitor that checks every 5 minutes can miss a 3-minute outage entirely. A 1-minute or 30-second interval gives a much truer picture.
- What counts as “up”. A server that returns an error page with a 200 status code can look healthy to a naive check. Good checks verify the status code, response time and sometimes page content.
- Where it’s checked from. A check inside your own network can pass while DNS or the CDN in front of you fails for everyone else. Measure from outside.
- Partial outages. Your homepage may load while login or checkout is broken. Monitor the pages and endpoints that matter, not just the front page.
How to measure your real uptime
Don’t rely on your host’s dashboard. It measures their infrastructure, not your visitors’ experience. Use an independent uptime monitor that checks your site from outside on a short interval and keeps a history you can report on.
What to look for:
- Short check intervals, 1 minute or less for anything that earns money.
- Confirmation before alerting. A failed check should be repeated before anyone is woken up, so a single network blip doesn’t trigger a false alarm.
- Alerts where you’ll see them: email, Slack, Discord, Telegram or a webhook, plus a recovery message with total downtime.
- SSL and domain checks alongside uptime, because expired certificates and domains are two of the most common self-inflicted outages.
- A public status page, so customers can see incidents and history instead of guessing.
FastPing, for example, re-checks a failure 20 seconds later before alerting, offers checks as often as every 30 seconds, and has a free plan with three monitors, enough to cover a small site, its certificate and its domain.
The takeaway
99.9% sounds perfect, but it allows almost nine hours of downtime a year. 99.99% allows under one, and costs far more to deliver. Decide what your site actually needs, remember that your real uptime is the product of everything it depends on, and measure it yourself. The number that matters isn’t the one in your host’s marketing. It’s the one your visitors experience.
