Uptime is the percentage of time your web hosting keeps your site reachable. “99.9% uptime”, the industry standard promise, allows about 43 minutes of downtime a month, or 8.8 hours a year, and that’s the ceiling before compensation kicks in, not a promise your site will stay up. Exclusions for maintenance, your own mistakes, and “circumstances beyond control” mean most real outages never qualify. And here’s what most guides skip: you don’t have to take the host’s word for any of it. You can monitor uptime yourself from day one.
On this page
The nines, translated
| Uptime | Downtime/day | Downtime/week | Downtime/month | Downtime/year |
|---|---|---|---|---|
| 99% (“one nine”) | 14.4 min | 1.7 hrs | 7.2 hrs | 3.65 days |
| 99.9% (“three nines”) | 1.4 min | 10.1 min | 43.2 min | 8.76 hrs |
| 99.95% | 43 sec | 5 min | 21.6 min | 4.38 hrs |
| 99.99% (“four nines”) | 8.6 sec | 1 min | 4.3 min | 52.6 min |
| 99.999% (“five nines”) | 0.9 sec | 6 sec | 26 sec | 5.3 min |
Each extra ‘nine’ cuts yearly downtime roughly tenfold. But costs disproportionately more.
Read it backwards: the jump from 99% to 99.9% looks like 0.9 percentage points and buys you 3.3 fewer days of yearly downtime. Each additional nine costs disproportionately more. Which is exactly why hosts advertise the nines and stay quiet about the price of the next one.
Why 99.9% isn’t what you think
It’s a ceiling, not a promise. The guarantee defines when compensation starts, not when your site stays up. A host at 99.5% against a 99.9% promise owes you a credit. Your visitors still got the outage.
It’s a percentage of what, exactly? Uptime is measured over a window, usually a month, and the window matters. Scheduled maintenance is typically excluded from the calculation entirely, so “99.9% excluding maintenance” can hide hours of planned downtime.
The exclusions list is where guarantees go to die. Standard carve-outs: scheduled and emergency maintenance, your own actions (bad plugin update, traffic spike you caused), third-party failures (CDN, payment gateway, DNS), and force majeure. Interpreted broadly, these cover most real-world outages. Which is why compensation is rare even when your site was genuinely down.
The compensation is credits, not cash. When a claim succeeds, you get account credit toward future service. Not a refund. And you usually have to file the claim yourself, in writing, within about 30 days, with timestamps.
The SLA claim playbook
If your site goes down and you suspect the guarantee was breached:
- Document immediately: start/end timestamps, what you observed (error type), and whether it was global or regional.
- Check the status page first: if the host posted maintenance, it’s excluded. Don’t waste the claim.
- File in writing within 30 days: support ticket or email, referencing the SLA section. Include your timestamps.
- Expect credits, not cash, and expect the host to determine at their sole discretion whether it qualifies.
- Track patterns: one credited outage is trivia; three in a quarter is a migration signal, regardless of what the SLA says.
Monitor it yourself. Stop taking their word for it
Set up independent uptime monitoring on day one. Your own record, not the host’s.
Competing guides usually stop at that admission. Here’s the missing step: set this up on day one with any new host:
- External uptime monitor: a free monitor that pings your site every 1-5 minutes from multiple locations and logs every outage with timestamps. This is your independent record. The one the host can’t edit.
- Alert on downtime: get notified within minutes, not when a customer emails you.
- Keep 90 days of history: that’s your evidence file for SLA claims and for deciding whether to renew or migrate.
- Monitor the checkout/signup flow, not just the homepage: a site can return “200 OK” on the homepage while the database-backed pages are dead. Monitor what earns money.
Cost: free tiers cover a single site comfortably. The 15 minutes of setup pays for itself the first time you file a claim, or the first time the data tells you to leave.
How much uptime do you actually need?
Match the nines to the stakes: blog, store, SaaS.
Not everyone needs five nines. Match the nines to the stakes:
| Your site | Target | Why |
|---|---|---|
| Personal blog / portfolio | 99.9% is plenty | An hour offline costs you nothing but pride |
| Business brochure site | 99.9% | Downtime is embarrassing, not expensive |
| Online store / booking site | 99.95%+ | Every hour offline has a revenue number. Calculate yours |
| SaaS / mission-critical | 99.99%+ | Downtime is churn; pay for redundancy, not just a promise |
The honest question isn’t “which host promises the most nines”. It’s “what does an hour of downtime cost me?” A blog chasing 99.99% is buying jewelry; a store accepting 99% is burning money.
Troubleshooting: uptime problems in the wild
- “My host claims 99.9% but my monitor shows 99.2%?” Check whether the outages fall in excluded categories (maintenance windows, your own plugin updates). If they’re unexplained server outages, file the SLA claim. With your monitor’s timestamps attached.
- “Site down but host says everything is fine?” Their monitoring checks the server; your monitor checks your site. A dead database or crashed PHP process looks “up” to them and down to visitors. Your monitor is the one that matters. If the downtime is a WordPress error rather than a server outage, investigate WordPress errors with our free troubleshooter.
- “Frequent short outages (2-5 min)?” These often come from resource limits (CPU throttling) rather than server failure. The “unlimited” fine print again. Check resource graphs before blaming the network.
- “Should I switch hosts over uptime?” Three unexplained outages in a quarter, or any single outage during your peak revenue hours with no honest post-mortem. That’s the migration threshold. When uptime keeps missing the guarantee, plan the move to a new host before the next outage costs you.
What to do next
- Choosing a host? Start with How to Choose Web Hosting: 7 Factors That Actually Matter.
- Choosing a tier? Read Shared vs VPS vs Cloud Hosting: Which Do You Actually Need?
- How we evaluate hosts: our How We Test page.
FAQ
Is 99.9% uptime good?
It’s the industry standard for shared and managed WordPress hosting. Acceptable for blogs and brochure sites. It still allows about 43 minutes of monthly downtime, so “good” depends on what downtime costs you.
What’s the difference between uptime and an uptime guarantee?
Uptime is what actually happened; the guarantee is the threshold that triggers compensation. A host can miss the lived experience badly while technically staying inside the guarantee’s exclusions.
Do hosts lie about uptime?
Rarely outright, but self-reported uptime excludes maintenance windows and is measured by their own systems. That’s why independent monitoring matters more than the marketed number.
Can I get compensation for downtime?
Sometimes: file a written claim within about 30 days with timestamps, and expect account credits rather than cash. Most real-world outages fall under exclusions, so successful claims are the exception. Which is exactly why your own monitor’s log is valuable.
Last updated: October 14, 2026. Plan terms change. Verify current SLAs before buying.