"99.9% uptime guaranteed" is on the homepage of almost every web host. It sounds like a strong promise. It isn't.
The Maths
Uptime percentages translate to allowed downtime per year:
| Uptime | Downtime per year | |---|---| | 99% | 3 days 15 hours | | 99.9% | 8 hours 46 minutes | | 99.95% | 4 hours 23 minutes | | 99.99% | 52 minutes |
99.9% — the most common claim — permits nearly nine hours of downtime annually. For a business whose customers can't reach their site during business hours, that's a significant amount.
What SLAs Actually Measure
The term "uptime" in a hosting SLA almost always means "the server is running and reachable." It typically excludes:
- Planned maintenance windows — scheduled downtime doesn't count against the SLA
- DDoS attacks — classified as external events outside the host's control
- DNS failures — if your DNS goes down but the server is up, usually excluded
- Third-party failures — payment processor, CDN, or upstream network issues
A host can have their server running 100% of the time while your site is unreachable for several hours due to a DNS misconfiguration or a DDoS mitigation measure — and none of that counts against their SLA.
What Credits Are Worth
Most hosting SLA credits are calculated as a fraction of your monthly fee. A typical formula: one day of service credit for each hour of downtime. For a £10/month plan, that's £0.33 per hour of downtime.
If your WooCommerce store processes £500/hour in orders, a three-hour outage costs £1,500 in lost sales. The SLA credit for that outage: £1. The credit is not compensation — it's a PR gesture.
What "Uptime" Doesn't Capture
A server can be technically "up" — responding to HTTP requests — while performing so badly that your site is functionally down. A time to first byte of 15 seconds is not an outage by any SLA definition, but it's not a functioning website for your visitors either.
Performance degradation on oversold servers happens constantly, affects real visitors, and is never reflected in uptime percentages.
How We Think About This
We don't advertise an uptime SLA figure because the industry's uptime theatre obscures more than it reveals.
What we do instead:
- Publish a live status page at
truecorehosting.com/status.html— updated every 60 seconds from real service checks - Give advance notice of all planned maintenance (minimum 48 hours for scheduled windows)
- Keep maintenance windows short — kernel updates and reboots complete in 2–5 minutes
- Build redundancy into DNS (three nameservers across separate locations) so DNS failures don't take your site offline
How to Monitor Your Own Site
Don't rely on your host's own uptime numbers. Run independent monitoring with a free third-party tool:
- UptimeRobot — free tier checks every 5 minutes, sends email/SMS alerts
- Freshping — free tier, 1-minute checks
- Better Uptime — free tier includes response time tracking
These send you an alert the moment your site becomes unreachable — from a server that has no relationship with your host. If your host claims 99.9% uptime but your monitor recorded three outages last month, you have the data to support that.
Independent monitoring also tells you what failed — HTTP response code, DNS timeout, TCP connection refused — which is more useful for debugging than a host's aggregated uptime percentage.