Website Uptime: What Your Hosting Provider's 99.9% Guarantee Actually Covers

Every managed WordPress hosting page on the internet has the same number on it. 99.9% uptime. Sometimes 99.95%. Occasionally 99.99% for premium tiers. Almost never explained.

That number is the single most-advertised operational claim in hosting. And somewhere between what it literally says and what most business owners think it says, there is a gap wide enough to drive a small business through.

Reviewing our incident response log over coffee this morning, I was struck by how rarely customers ever ask about uptime in advance. They ask after a site has gone dark, usually on a Saturday evening, usually when a customer or a lead has just told them the website is broken. By then the conversation is not about percentages. It is about how long, why, and what happens next.

This is a plain-English explainer of what your hosting provider's uptime number actually guarantees, what it quietly does not cover, what you can realistically claim when it fails, and how to verify it yourself. No jargon unless I define it first.

What 99.9% uptime actually means in hours

Availability percentages are multiplication tricks. They translate into real time like this, and the jump between tiers is bigger than most people assume.

Uptime advertised	Max permitted downtime per year	Per month
99%	Around 87 hours (3.6 days)	About 7 hours
99.9%	Around 8 hours 45 minutes	About 43 minutes
99.95%	Around 4 hours 22 minutes	About 21 minutes
99.99%	Around 52 minutes	About 4 minutes
99.999%	Around 5 minutes	Roughly 26 seconds

Figures confirmed against the standard SLA calculator at uptime.is and cross-checked with Wikipedia's high-availability reference tables ^[1][2].

So a 99.9% uptime guarantee is not "basically always on". It is "permitted to be offline for almost nine hours in the calendar year, and your provider is still meeting their contract". The step from three nines to four nines is roughly an order of magnitude. The step from four to five is another. Most shared and lower-tier managed plans are sold at three nines.

Nine hours is long enough to contain every high-stakes window your business has. A Black Friday weekend. A payroll Friday. The dinner-booking rush on a bank holiday Sunday. You can technically meet 99.9% while being offline for every moment that actually mattered to the business.

The five things your uptime number does not cover

The headline figure is the cleanest part of a hosting SLA. The exclusions do most of the work, and most of them are industry standard.

1. Scheduled maintenance. Most SLAs exclude any downtime during a notified maintenance window. WP Engine's public Service Level Agreement is explicit about this, and most competitors word it the same way ^[3]. Your provider can take the site offline for several hours with advance notice and still show 100% uptime on their dashboard.
2. Emergency maintenance. Usually defined as unplanned work "reasonably required" to protect security or integrity. Also excluded.
3. Causes beyond reasonable control. Force majeure, upstream network incidents, DDoS attacks, third-party service failures. This is the widest category and the one most often invoked when real outages happen.
4. Customer-caused issues. Plugin conflicts, a bad theme update, a misconfigured caching rule, a runaway WooCommerce query. If the provider can point at your side of the stack, the downtime does not count.
5. Traffic spikes. Many plans explicitly exclude downtime caused by exceeding your resource allocation, even if no one warned you that you were about to hit it.

An outage that feels catastrophic to a business owner, say four hours offline on a busy Saturday, can still be "within SLA" to the provider because one of those clauses applied. That is not a hosting problem. That is a contract design problem, and it is identical across most of the industry.

What actually happens when a site goes down

There are two clocks running during an outage, and they do not tick in sync.

The technical clock is short. A server returns 5xx errors. Monitoring detects it in anywhere from 30 seconds to several minutes depending on check frequency. An on-call engineer responds. The fix is deployed. The site comes back.

The business clock is longer. Each minute the site is unreachable, a visitor trying to buy, book or enquire either waits, gives up, or goes to a competitor. Research on this is all over the map. The common range for small business downtime cost sits somewhere between roughly €120 and €400 per minute once you include lost revenue, lost leads, and brand impact, according to Atlassian's methodology write-up and several industry studies ^[4]. I would take those figures with a caveat. Methodology varies by sector, and a brochure site for a small consultancy sees a different number than a WooCommerce shop mid-checkout. But the direction is right.

Then there is the third clock, which ticks for weeks. Google's own guidance, as reported by Search Engine Journal, is that short intermittent downtime does not affect rankings, and that is true ^[5]. But sustained outages, days rather than minutes, reduce crawl budget and can push indexed pages out of search for longer than the outage itself. A one-afternoon blip is fine. Three outages across a month can quietly cost you positions before you have noticed anything has changed.

A boutique hotel owner on the Wild Atlantic Way does not care about crawl budget at 8pm on a Saturday. They care that booking enquiries stop coming in and the listing one further down on Google picks up the slack. That is where uptime stops being a number and starts being a business decision.

The credit you cannot actually claim

Here is the part of SLAs most people have never read. When your provider misses their uptime commitment, the remedy is almost always a service credit on your next invoice, not a refund and not compensation for business loss.

The arithmetic is comical. If you pay around €100 a year for shared hosting and your site is offline for a full day, the typical SLA credit works out to single-digit euros. Liquid Web, which runs one of the more generous models in the industry, credits 10x the downtime duration as account credit, so one hour of outage earns ten hours of hosting credit ^[3]. That is still not cash. It is a discount on a future month with the same provider who just let you down.

The credit is also never automatic. You usually have to file a claim within a specified window, often 30 days, with your own documentation of the outage. Most business owners never file. They either do not know about the process or decide it is not worth an afternoon of paperwork for €5.

Do not base your hosting decision on the headline percentage. The SLA is a marketing document in most of its length and a legal document in the fine print. The real question is not "what number do they advertise" but "what do they do operationally to avoid needing to invoke the SLA in the first place".

How to measure uptime yourself

You do not have to take your provider's word for any of this. Independent uptime monitoring is cheap and, in many cases, free. UptimeRobot, StatusCake and Pingdom all offer plans that will check your site every minute or two from multiple geographic locations and alert you within seconds of an outage ^[3].

A few practical notes:

• Monitor from at least two regions. A check from one node can misreport if that node itself has a problem.
• Set the check interval to one minute if your hosting plan allows it. Five-minute intervals can miss short outages entirely.
• Have alerts go to at least two people, and at least one channel that is not email. If your email uses the same domain as the site being monitored and the email server is also affected, you will never see the alert.
• Keep the monthly uptime report. Do not trust a provider's own status page as evidence of their own downtime.

When something goes wrong, you now have independent data. You know when it started, how long it lasted, and whether the provider's dashboard agrees with reality. That is the difference between a customer with a complaint and a customer with a case.

The stakes by business type

Uptime is not one problem. It is three problems depending on how your site earns its keep.

• eCommerce and WooCommerce. Downtime is lost revenue in real time. A cart abandoned mid-checkout rarely returns. If payment has already been authorised but not captured, you inherit a customer service problem on top of the sale you lost.
• Lead generation and service businesses. Downtime is lost enquiries you will never know you missed. The form that did not load, the phone number that was never seen, the quote request that did not go through. The cost shows up later as a quieter month.
• Hospitality and bookings. Downtime during peak search windows, evenings and weekends, is directly competitive. A visitor who cannot see your menu or your rooms picks the next listing on Google within ninety seconds.

Each of those needs the same things from hosting. A proper stack. Real monitoring. An incident response process that does not depend on a single overseas support queue. But the consequence of getting it wrong is different, and so is the business case for paying attention.

What proper uptime looks like on the operations side

Strong uptime is not one feature. It is a stack.

Underneath it sits the infrastructure layer. Redundant power, redundant network, data centres that are not sharing fate with an outage somewhere in Virginia. On top of that, a hosting stack built for WordPress specifically, with Nginx, PHP-FPM, Redis object caching and FastCGI page caching configured by people who do WordPress for a living. Above that, verified nightly backups, staging environments to catch updates before they break production, and a monitoring layer that wakes someone up before a customer notices.

That is the picture of what Web60's Irish-infrastructure managed WordPress hosting delivers for €60 a year all inclusive. The hosting runs on SmartHost's sovereign Irish cloud. The stack is WordPress-native rather than generic. Backups are automatic, nightly, verified, and restorable in one click. Support is handled by people in an Irish timezone. The uptime number we publish is the output of that setup, not a marketing figure bolted onto a cheaper stack.

For a deeper walk-through of how the caching layers connect together, the WordPress performance stack and its caching layers piece breaks it down in detail. For the broader performance picture the stack is trying to deliver, the complete WordPress performance guide for business owners is the pillar reference.

The strategic concession

If your website is a true brochure site with no revenue dependency, a handful of visits a week, and you will genuinely not mind if it is offline for an afternoon, then any mainstream hosting plan with a 99.9% promise is probably fine for you. The uptime conversation only starts to bite when the website is part of how the business makes money. For most local firms reading this, it is.

A lesson I had to learn properly

A few years ago I trusted a monitoring alert that said a site was fine because the homepage responded 200 OK. The homepage was static, served from the cache. The checkout endpoint was throwing 500s and the cache had shielded us from noticing for nearly 40 minutes. We changed the monitoring approach the next day. Check the pages that make money, not the pages that look good. Lesson paid for.

Sync reality check

Uptime monitoring is not prevention. It tells you your site has gone down, faster than a customer, and it gives you evidence after the fact. It will not stop an outage the way good architecture does. A well-monitored cheap shared host will still go down at the same frequency as an unmonitored one. Monitoring shortens recovery, not causes. Know the difference when you are shopping.

Conclusion

The uptime percentage on a hosting landing page is where the conversation starts, not where it ends. The honest question to ask any provider is not "what is your uptime guarantee" but "what is your scheduled maintenance window policy, what is excluded from your SLA, what credit do I actually get, and when was your last unplanned outage". Any provider that answers those four questions clearly is worth considering. Any provider that only points at the big number on the homepage is not.

From your own side of it, decide how much downtime your business can genuinely absorb and in which windows. Monitor independently. Read the exclusions. And pick a stack that treats uptime as an operational practice rather than a marketing number.

Frequently Asked Questions

Is 99.9% uptime good enough for a small business website?

99.9% is the industry floor, not the ceiling. It permits roughly 8 hours 45 minutes of downtime per year. For a brochure site, that is usually fine. For an eCommerce site, or one that takes bookings at weekends, it is worth looking at 99.95% or better and at what the provider does operationally to earn that number.

Does website downtime hurt Google rankings?

Short intermittent downtime does not, according to Google's own guidance. Sustained outages spanning days can reduce crawl budget and drop indexed pages, with recovery taking a couple of weeks once the site is stable again.

What is excluded from most hosting SLAs?

Scheduled maintenance, emergency maintenance, force majeure, third-party service failures, customer-caused issues (plugin conflicts, bad updates), and traffic exceeding plan limits. These exclusions cover most real-world outages.

Can I claim compensation if my website goes down?

In most cases you receive a service credit toward your next invoice, not a refund. You usually have to file a claim within a defined window, and the credit often amounts to a few euros even for a full day of downtime. Read your SLA before it matters.

How should I monitor my own website's uptime?

Use an independent service such as UptimeRobot, StatusCake or Pingdom. Monitor from at least two geographic regions, check every one to two minutes, and send alerts to at least two people on a channel that does not share infrastructure with your website.

Does Web60's hosting include uptime monitoring?

Yes. Web60 includes platform-level monitoring, nightly backups, staging environments and Irish-based support as part of the €60 a year plan. Adding independent third-party monitoring on top is still worth doing, so you have a record the provider cannot rewrite.

Sources