The true cost of downtime: What an hour offline actually costs your business

31 Aug 2026, by Micron21

Across almost every article we write, there's one word that keeps coming up: uptime.  Whether we're talking about our Tier IV data centre, high availability (HA) architecture, redundant networking, or the Service Level Agreements (SLAs) that back our services, the underlying promise is always the same - keeping your systems online and available.  But whilst we talk about “uptime” constantly, there's one thing we've never actually done - and that is to put a dollar figure on the alternative of not having uptime, which is called “downtime”.

That's a strange omission, because downtime is one of those costs that's easy to underestimate right up until the moment it happens.  When your systems go offline, there's no invoice that lands on your desk, but the meter starts running all the same.  Staff sit idle, customers can't buy, orders stall, and the costs quietly pile up in the background whilst everyone scrambles to get things working again.

And it's a problem that's only becoming more expensive - as businesses of today become more dependent on their digital systems than ever before.  Nowadays, outages remain remarkably common.  Research suggests that around 80% of data centres have experienced at least one outage in the past three years¹ and the per-hour cost of that downtime keeps climbing year on year.

That's why in this month's blog we'll be quantifying it.  We'll break an hour of downtime down into real dollars across lost revenue, lost productivity, recovery, and reputation.  We’ll show how even "good" uptime figures can hide more downtime than you'd think, and we’ll also explain why uptime is ultimately a business decision rather than just a technical one.

Before we start attaching numbers to it, it's worth being clear about what "downtime" actually covers - because it's broader than most people assume.

At its simplest, downtime is any period where your systems, applications, or services are unavailable to the people who rely on them - whether that's your customers, your staff, or both.

It comes in a few different flavours.  There's planned downtime (scheduled maintenance windows, upgrades, migrations) and unplanned downtime (hardware failure, power loss, network outages, cyberattacks).  Then you have total downtime where everything goes dark, compared to partial downtime where a single component fails (imagine your website staying up but your checkout or payment gateway going offline).  And then there's a distinction between infrastructure-level downtime (the server, power, or network underneath) and application-level downtime (the software running on top).

The important point here is that not all downtime looks like a dramatic, lights-out blackout scenario.  However, in saying that, degraded performance - such as a site that is crawling or an application that keeps timing out – also carry a cost too, even though technically "everything's still up".   To illustrate this, customers trying to complete a purchase at a checkout that is taking way too long to load, is often perceived just as bad as a checkout that's completely offline.

For those who want the infrastructure-level breakdown of why  genuinely high uptime is so difficult (and expensive) to achieve, we'd recommend our Deep Dive – What is Tier IV? article, which explains the redundancy that sits behind it.

So, what does downtime actually cost?  The honest answer is "it depends" - but that's not very satisfying, so let's break it down into its component parts.  Broadly, the cost of an outage falls into four buckets:

1. Lost Revenue:  This is the most obvious cost, and often the largest.  It's every sale, transaction, and booking that simply doesn't happen whilst you're offline.  For an e-commerce store, an online retailer, or any organisation that takes orders or payments through its systems, this is money walking straight out the door.  And timing matters enormously - an outage on a quiet Tuesday night is a very different proposition to one that hits during a major sale, a product launch, or your busiest trading hour of the year.

2. Lost Productivity:  Even if you don't sell anything online, downtime still costs you.  When your systems are down, your staff often can't do their jobs - but the wage bill keeps running regardless. Every employee sitting idle, waiting for systems to come back, is being paid to produce nothing.  Across a whole team, for even a few hours, that adds up quickly.

3. Recovery Costs:  Getting back online is rarely free.  There's the overtime for your own team, the emergency call-out fees for external contractors or vendors, the cost of restoring data from backups, and all the cleanup work that follows once the immediate crisis has passed.  In the worst cases – such as a ransomware attack, or a serious hardware failure - recovery can stretch on for days rather than hours.

4. Reputation and Trust:  This is the slowest-moving and hardest-to-measure cost, and often the one that lingers longest.  Customers who can't reach you don't always come back.  An outage that makes the news, or that frustrates enough people on social media, can erode trust that took years to build.  And for businesses with contractual uptime commitments of their own, downtime can also mean SLA penalties, refunds, or lost future deals.  You won't see this cost on a spreadsheet the day it happens - but you'll feel it in the months that follow.

To make this a little more concrete, let's run some rough numbers.  Take a business turning over $10 million a year.  Spread across a standard working year, that's roughly $192,000 in revenue per week, or a little over $4,800 for every working hour - and that's just the lost-revenue component, before you add a single dollar of lost productivity or recovery costs on top.  Then on top of this, layer in idle staff plus the cost of putting things right, and you start seeing the true figure climbing even higher still.  Suddenly, "just a couple of hours offline" starts to look a lot more expensive than it sounds.

And that's a fairly modest example.  When you look at the broader research, the numbers can be genuinely sobering:

• In ITIC's 2024 Hourly Cost of Downtime Survey, over 90% of mid-to-large enterprises reported that a single hour of downtime would cost them more than $300,000 - and around 41% put the figure between $1 million and $5 million (or more) per hour.²
• A long-standing Gartner benchmark placed the average cost of downtime at roughly $5,600 per minute - about $336,000 an hour - and most research since suggests that figure has only trended upwards.³
• New Relic's 2025 Observability Forecast pegged the median cost of a high-impact outage at around $2 million per hour.
• Smaller organisations aren't exempt either.  Industry analysis of the ITIC figures suggests that even micro-businesses (under 25 staff) can face costs in the vicinity of $100,000 per hour, whilst businesses with 20 to 100 staff commonly land somewhere between $8,000 and $25,000 per hour.²

It's also worth remembering that outages aren't always over in a few minutes.  Whilst the median significant outage runs for somewhere around 53 minutes, there's a long tail - a meaningful share stretch beyond four hours, and some drag on for more than a full day.  Multiply any of the hourly figures above by a whole day, and the case for investing in uptime rather tends to make itself.

Whenever downtime comes up, there's a natural and very human tendency to assume it's someone else's problem.  "We've got backups."  "We're in the cloud."  "We're too small for it to matter."   These are the three reassurances we hear most often, but unfortunately, each one has a catch.

"We have backups."  Backups are essential, so that's great - but if you don't have them properly isolated and regularly tested, then that's a totally different conversation worth having on its own.  What we're talking about here is that it's important to understand what backups actually protect – which is your data, not your uptime.  A backup means you can recover after an outage - it does nothing to prevent the outage itself.  In fact, restoring from backup is itself a period of downtime, and depending on how much data you're moving, that restore can take hours or longer.  So backups are your safety net - not your uptime strategy.

"We're in the cloud."  Moving to the cloud doesn't automatically make you resilient.  This is one of the most common misconceptions we come across.  Cloud infrastructure still runs on physical hardware in physical data centres, and it can and does fail.  A single availability zone going down, or a provider-wide outage, can take your services offline just as effectively as a server dying in a cupboard down the hall.  What actually determines your resilience is how your environment is architected - whether there's genuine redundancy and failover built in, or whether you've simply moved a single point of failure from your office to someone else's.

"We're too small for it to matter."  It's true that the eye-watering  per-hour figures quoted earlier belong mostly to large enterprises.  But whilst the absolute dollars are smaller for a small or medium business, the proportional  impact is often far worse.  A larger organisation can usually absorb a bad day.  However, for a smaller business, a single serious outage can wipe out an entire month's margin - or worse!  Smaller doesn't necessarily mean safer.

This is also where it pays to look closely at what an uptime figure actually means  in practice - because the numbers can be deceiving. Uptime is usually expressed as a percentage, and small-looking differences translate into very different amounts of real-world downtime over the course of a year:

• 99.9% uptime ("three nines") sounds excellent - but it still allows for roughly 8.75 hours of downtime per year.  That's most of a working day, offline, every single year.
• 99.99% uptime ("four nines") brings that down to around 53 minutes per year.
• 99.995% uptime cuts it further, to roughly 26 minutes per year.
• 100% uptime SLA, backed by genuinely fault-tolerant (2N+1) Tier IV infrastructure, is designed to eliminate that gap entirely - with no single point of failure to bring things down in the first place.

Putting it another way, the difference in uptime between "three nines" and a 100% SLA isn't a rounding error.  It's the difference between quietly accepting the better part of a working day offline each year and designing things so that it doesn't happen at all.

This is exactly where the infrastructure we talk about so often comes into its own.  Our Tier IV data centre, high availability (HA) architecture, and SLA-backed services are all, ultimately, about the same thing: reducing both the likelihood of downtime and how long it lasts when something does go wrong.  When you weigh the cost of that resilience against the per-hour figures above, it tends to look less like an expense and more like insurance with a very favourable premium.

Earlier we made the point that being "in the cloud" doesn't automatically make you resilient - and that it all comes down to how the platform is actually built underneath.  That's a fair warning about cloud in general, but it's also exactly the gap we set out to close when we built mCloud.  Put simply, mCloud is designed to be as resilient as people usually assume all clouds should be and has been built to this expectation, because those others often don't provide it.

It all starts with where it lives.  mCloud runs inside our 100% owned and operated Australian Tier IV-certified data centre, which means the fault-tolerant (2N+1) power, cooling, and networking we've talked about throughout this article sit underneath every workload by default - not as an optional extra that you have to bolt on later.

From there, we build in redundancy and replication at the platform level rather than leaving it to chance. Rather than keeping your data confined to a single location, mCloud synchronises it across three geographically separate data centres, so a problem in any one site doesn't take your systems down with it. Our storage layer - our NVMe-powered mSAN platform, built on Ceph - is engineered for exactly this, with real-time replication across locations and no single point of failure to bring things to a halt.  In practice, that's the difference between a single hardware fault being a genuine disaster and it being a non-event your workloads simply ride straight through.

It's also why high availability (HA) comes as standard rather than as a premium tier.  Built on OpenStack and Ceph - which are proven, enterprise-grade open-source technologies - mCloud is designed with out-of-the-box high availability (HA), giving you the kind of resilience normally associated with the global hyperscalers, but hosted here in Australia, backed by our 100% uptime SLA, and supported around the clock by an Australian-based team.  When you tie that back to the cost of downtime, the value proposition is straightforward - that is, that the most effective way to avoid paying for an outage is to build on a platform designed so the outage never reaches you in the first place.

If you have any questions about what downtime could really cost your business - or you'd like to talk through how to reduce your exposure to it - let us know!  We're always happy to take a look at your setup and walk through the options with you.

We can help with Tier IV colocation, high availability (HA) architecture, disaster recovery, and SLA-backed hosting - all designed to keep you online when it matters most.

You can reach us via email at sales@micron21.com or call us on 1300 769 972 (Option #1).

1, Uptime Institute, Annual Outage Analysis (data centres experiencing an outage in the past three years) - as reported by Gatling, "The cost of downtime", <https://gatling.io/blog/the-cost-of-downtime>
2, ITIC (Information Technology Intelligence Consulting), "2024 Hourly Cost of Downtime Survey" - as reported by MEV, "The Cost of IT Downtime in 2025: What SMBs Need to Know", <https://mev.com/blog/the-cost-of-it-downtime-in-2025-what-smbs-need-to-know>
3, Gartner, cost-of-downtime benchmark (~$5,600 per minute) - as reported by Virima, "IT Downtime Cost Statistics: What Outages Cost in 2025", <https://virima.com/blog/it-downtime-cost-statistics-what-outages-actually-cost-in-2025>
4, New Relic, "2025 Observability Forecast" (median high-impact outage cost) - as reported by Virima, "IT Downtime Cost Statistics: What Outages Cost in 2025", <https://virima.com/blog/it-downtime-cost-statistics-what-outages-actually-cost-in-2025>
5, Uptime Institute, "2025 Global Data Center Survey" (median outage duration and long-tail figures) - as reported by JustAnalytics, "The Cost of Downtime in 2026", <https://justanalytics.app/blog/cost-of-downtime-statistics-2026>