Every WordPress host advertises an uptime guarantee, and the number they advertise tells you almost nothing. What matters is the remedy: what the host owes you when they miss it, how long you have to claim it, and what is quietly excluded. Those terms vary far more than the headline percentage does.
We read the published service level agreements of seven WordPress hosts and worked out what each one actually pays for the same outage. On an identical plan, a single hour of downtime is worth $0.42 at one host and $6.00 at another. The host advertising the most aggressive sounding guarantee in the industry pays the least of any of them.
- Three of the seven hosts advertise 100% uptime. They pay wildly different amounts when they miss it, so the headline number is close to meaningless on its own.
- For the same one-hour outage on the same plan price, the credit ranges from $0.42 to $6.00, a 14x spread.
- Liquid Web’s “10 times the downtime” credit sounds like the most generous term in hosting. Do the arithmetic and it is the smallest payout in this table.
- Every single SLA is opt-in. No host credits you automatically. If you did not notice the outage, you get nothing.
- Claim windows are short and vary a lot: 5 business days at Liquid Web, 7 at Nexcess, 14 at DreamHost, 30 at WP Engine, Kinsta and Pressable.
- Rocket.net publishes no uptime SLA at all in its terms of service.
What each host actually promises
These figures come from each company’s own published SLA or terms of service, read in August 2026. Hosts are included here when they publish a retrievable agreement with specific remedy terms.
| Host | Guarantee | What you get when they miss it | Cap | Claim window |
|---|---|---|---|---|
| WP Engine | 99.95% | 5% of monthly fees per full hour of excess downtime | 100% of monthly fees | 30 days |
| Kinsta | 99.9% | Banded: 5% (43 to 59 min), 10% (60 to 119), 15% (120 to 179), 20% (180 to 239) | Monthly subscription value | 30 days, in writing |
| Pressable | 100% | 5% of monthly fee per 30 min of network or infrastructure downtime | 100% of monthly fee | 30 days |
| Nexcess | 100% | 5% of monthly fees per additional 15 min | Monthly fees for affected service | 7 days |
| DreamHost | 100% | One day of hosting cost per hour (or partial hour) of interruption | 10% of next renewal | 14 days |
| Liquid Web | 100% network | 10 times (1,000%) the actual downtime, as service credit | Monthly recurring charge | 5 business days |
| Rocket.net | No uptime percentage or credit formula in its published terms of service | |||
Two things jump out before you even do the maths. Three hosts advertise 100%, which cannot literally mean never down, and in practice means “we start owing you from the first minute” rather than “we allow you 43 minutes first.” And DreamHost is the only host here that caps the credit at a fraction of one bill, at 10% of your next renewal, no matter how long the outage ran.
What the percentage actually allows
Before the remedy matters, the guarantee has to be missed. Here is how much downtime each level permits before a host owes you anything. A 30-day month is used, which is the convention most SLAs follow.
| Guarantee | Allowed per month | Allowed per year |
|---|---|---|
| 99% | 7h 12m | 87h 36m |
| 99.5% | 3h 36m | 43h 48m |
| 99.9% (three nines) | 43m 12s | 8h 45m |
| 99.95% | 21m 36s | 4h 22m |
| 99.99% (four nines) | 4m 19s | 52m 34s |
| 99.999% (five nines) | 26s | 5m 15s |
This is where the gap between 99.9% and 99.95% stops being cosmetic. Moving up one step halves the downtime a host may take before owing you anything, from 43 minutes a month to 21. WP Engine’s 99.95% is a meaningfully stricter commitment than the 99.9% that most of the industry advertises, and it is the kind of detail that gets flattened when a comparison post rounds everything to “99.9% or better.”
One caveat worth knowing if you ever check a host’s maths against your own: some SLAs measure a month as 30 days and some use the actual calendar month. At 99.9% that is the difference between 43m 12s and 43m 50s. It rarely matters, but it explains why two calculators disagree slightly.
What a one-hour outage actually pays you
Comparing “5% per hour” against “10 times the downtime” against “one day of hosting” is impossible in the abstract, so here is every formula applied to the same event: a single hour of downtime on a $30 per month plan. The plan price is hypothetical and used only so the formulas compare like for like.
| Host | Formula | Credit for one hour down |
|---|---|---|
| Nexcess | 5% per 15 min, so 20% | $6.00 |
| Kinsta | 10% band for 60 to 119 min | $3.00 |
| Pressable | 5% per 30 min, so 10% | $3.00 |
| WP Engine | 5% per full hour | $1.50 |
| DreamHost | One day of hosting per hour | $1.00 |
| Liquid Web | 10x downtime, so 10 hours of service | $0.42 |
A 14x spread between the best and worst outcome, from terms that all sound broadly similar when a host summarises them on a features page.
The 1,000% credit that pays the least
Liquid Web’s SLA is widely described as the most aggressive guarantee in hosting, and the wording invites that reading: you receive “a credit for 10 times (1,000%) the actual amount of downtime.” Their own example is that one hour down earns ten hours of credit.
Ten hours of what, though. Ten hours of service. A $30 monthly plan contains roughly 720 hours, so an hour of it is worth about four cents, and ten hours is worth about forty. The “1,000%” is a multiple of the downtime, not a multiple of your bill, and those are very different things.
Nexcess, which is part of the same company, uses the plainer sounding “5% of monthly fees per additional 15 minutes.” That works out to 20% of the month for the same hour, roughly fourteen times more money. The percentage that sounds enormous is a multiplier on a tiny base, and the percentage that sounds modest is a multiplier on your actual bill. This is the single most useful thing to understand when reading any hosting SLA.
Every one of these is opt-in
Here is the part that makes most of the above academic. Not one of these hosts credits you automatically. Every agreement requires you to notice the outage, report it, and ask, inside a fixed window:
Liquid Web gives you the least room, at five business days from the incident. Nexcess wants a report within seven days of the outage ending, filed as a ticket. DreamHost allows fourteen, with documentation. WP Engine, Kinsta and Pressable are the most forgiving at thirty, though Kinsta’s terms add that failing to ask in writing “will forfeit Customer’s right to receive an SLA Credit.”
Five business days is a genuinely tight window. A short outage at the weekend, on a site you are not watching, can expire before you know it happened. Hosts do not email you to say they missed their own SLA.
Which means the guarantee is only worth anything if you have independent evidence of when your site was down. Your host’s own status page is their account of events, not yours. If you are not monitoring, you have no claim, and you also have no way to know whether the 99.9% you are paying for is being delivered. You can check whether your site is up right now with our free uptime checker, and any of the established monitoring services will keep a dated log you can attach to a support ticket.
What the exclusions quietly remove
Every SLA here carves out categories of downtime that do not count. The usual ones are reasonable: force majeure, your own misconfiguration, breaching the acceptable use policy. A few are worth reading twice.
- Scheduled maintenance is excluded everywhere. Kinsta defines a daily window between 2am and 5am, and downtime inside a declared window is not downtime for SLA purposes, even though your visitors experience it identically.
- Liquid Web excludes malicious attacks and DDoS, which is one of the more common reasons a site becomes unreachable in the first place.
- At Kinsta, traffic beyond your plan’s capacity is excluded. Going down because you got popular is on you.
- Credits usually expire with the account. WP Engine states they are not carried over or aggregated and are forfeited when the agreement ends, so one earned on your way out is worth nothing.
- The credit is normally the only remedy. Pressable calls its guarantee “your sole and exclusive remedy,” standard language across the industry, which means whatever the outage cost your business, the credit is the ceiling.
What to actually do with this
Do not pick a host on its uptime percentage. Nearly everyone advertises 99.9% or better, and the number is a marketing input more than an engineering commitment. Three of the seven hosts here claim 100%, which no infrastructure achieves indefinitely.
What the SLA is genuinely useful for is as a signal of confidence. A host that publishes specific, banded, generous remedies is telling you it does not expect to pay them often. A host that publishes no SLA at all, or caps the remedy at a token fraction of one bill, is telling you something too.
Three practical moves:
- Monitor independently, because without your own record you cannot claim, and cannot tell whether the guarantee is being honoured at all.
- Write your claim window down now, before you need it. Five business days is a calendar reminder, not something to go looking for in the middle of an outage.
- Size the credit against the damage. If an hour offline costs you more than a few dollars, and for most commercial sites it does, the SLA is not insurance but a gesture. Budget for backups, staging and a migration plan rather than leaning on the guarantee.
If you are weighing hosts more broadly, our comparison of the best WordPress hosting covers performance, support and pricing alongside these guarantees.
Frequently asked questions
How much downtime does a 99.9% uptime guarantee allow?
43 minutes and 12 seconds per month, on a 30-day month, and 8 hours 45 minutes per year. Your host can be down for that long and still be meeting its guarantee, so no credit is owed. At 99.95% it halves to 21 minutes 36 seconds a month, and at 99.99% it drops to 4 minutes 19 seconds.
Can a host really guarantee 100% uptime?
Not in the sense of never going offline. A 100% guarantee means the host starts owing you a credit from the first minute of unplanned downtime rather than allowing a monthly allowance first. It is a billing commitment, not a physics claim, and scheduled maintenance is still excluded. Pressable, Nexcess and DreamHost all advertise 100% in this sense.
Do hosts pay SLA credits automatically?
No. Every agreement reviewed here requires the customer to report the outage and request the credit within a set window, ranging from 5 business days to 30 days. Kinsta’s terms state explicitly that failing to submit a written request forfeits the right to the credit. If you do not notice an outage, you will not be compensated for it.
Is an SLA credit worth switching hosts for?
On its own, no. Even the most generous formula here returns a few dollars for an hour offline, and every agreement makes the credit the sole remedy, so it caps what you can recover. Treat the SLA as a signal of how confident a host is, then choose on performance, support and architecture.
Does scheduled maintenance count against the guarantee?
No, at every host reviewed. Announced maintenance windows are excluded from the uptime calculation, so time your site spends unreachable during one earns no credit. Kinsta publishes a daily 2am to 5am window; DreamHost excludes maintenance announced at least five days ahead and not exceeding one hour per month.
Bottom line
The advertised percentage is the least informative part of a hosting uptime guarantee. The remedy, the cap and the claim window are where hosts actually differ, and they differ by more than an order of magnitude for identical downtime.
The practical conclusion is unglamorous: no SLA in this table will meaningfully compensate a commercial site for being offline. What the guarantee is good for is telling you how confident a host is in its own infrastructure, and giving you something to point at when an outage runs long. Both of those require you to have your own record of what happened, which is the one part of this entirely within your control.
