Every host puts an uptime number on the pricing page. 99.9%, 99.95%, 99.99%. They all look close enough to 100 that you skim past them.
Two things are worth understanding. What those decimal places cost you in real hours, and what the guarantee behind them actually covers. The second one is the part almost nobody reads, and it is where the money is.
The decimals in hours
Percentages hide the scale. Convert them to time and they stop being abstract.
- 99.9% allows about 8.7 hours of downtime per year. Around 43 minutes a month.
- 99.95% allows about 4.4 hours per year.
- 99.99% allows about 53 minutes per year.
- 99%, which some cheap hosts still quote, allows about 3.65 days per year.
So the gap between 99.9% and 99.99% is roughly eight hours a year. Whether that matters depends entirely on what an hour of your site being down is worth. For a brochure site, it is a rounding error. For a store doing $4,000 a day, eight hours is $1,300 and a pile of support emails.
The fix: work out your own number. Monthly revenue through the site, divided by 730 hours. That is what an hour of downtime costs you, and it tells you how much uptime is worth buying.
The exclusions are the whole story
Here is the part that surprises people. A host's uptime SLA covers the server. Only the server.
Read the terms and you will find the carve-outs written plainly: application errors, third-party plugins and themes, customer-installed software, database issues caused by software rather than hardware, and anything the customer or their developer did. Scheduled maintenance is usually excluded too.
Now think about how WordPress sites actually go down. In practice it is almost never the hardware. It is a plugin update that conflicts with the theme. It is a PHP fatal error after an automatic update. It is malware injected through an outdated plugin. It is a database that fills up or a memory limit that gets hit at checkout.
Every one of those is excluded from the uptime guarantee. Your site is down, your customers cannot buy, and your host's uptime for the month is still 100% because their server responded to every request. They are not lying. They are measuring something different from what you care about.
What the credit is worth
If a host does breach its SLA, you get a service credit. Note the word credit, not refund. It comes off your next invoice.
On a $30 a month plan, a typical breach credit works out at a few dollars. If that outage cost you a day of sales, a $3 credit is not compensation, it is a gesture. Some hosts go further and reserve the right to decide whether a breach occurred at all, at their discretion, which is a guarantee in the same way a coin flip is a plan.
None of this means uptime SLAs are worthless. It means you should read one as a statement about infrastructure quality, not as insurance for your business.
The fix: before you take an uptime number seriously, find the exclusions section and the credit calculation. Five minutes of reading tells you more than the number on the pricing page.
Uptime SLA versus care SLA
This is the distinction worth carrying away.
An uptime SLA is a promise about a machine. The server will respond. If it does not, here is a credit.
A care SLA is a promise about your site. If it breaks, somebody fixes it. Not alerts you, not opens a ticket about it, fixes it.
Those two promises cover almost opposite failure modes. Hardware is reliable. Software is not. Which means the promise most WordPress owners actually need is the one hosts do not make.
What we commit to, and what it covers
Up Speed commits to 99.9% uptime, monitored 24/7 with checks every 60 seconds. That is about 8.7 hours of unplanned downtime a year at the outside, and we quote 99.9% rather than a higher number because it is the one we can stand behind. If we breach it in a billing month, you get a service credit.
The credit is not the point. The coverage is. Our SLA is a care SLA, so plugin and theme conflicts, malware, failed updates and performance regressions are inside the scope, not carved out of it. Updates get tested before they go live and rolled back when one causes a problem. Backups run off-site on a schedule with one-click restore. The commitment is to get your site working again, not just to keep a server pinging.
It is honest scope, not an unlimited promise. Things genuinely outside our control, like your domain registrar going down or a paid API you rely on failing, stay outside it, and direct code changes we were not told about may be billable to untangle.
If an hour of downtime costs you real money, the response time is the number to look at. The Premium plan carries a 1-hour emergency response SLA and a dedicated point of contact who knows your site, so there is no explaining your setup to a stranger at the worst possible moment.