What one hour of downtime actually costs a small business

When a system your business runs on goes down — the point-of-sale, the shared drive, the line-of-business app, the internet connection itself — the meter starts running the moment it stops. Most owners have a vague sense that downtime is expensive. Far fewer have ever put a number on their own. That number is worth knowing, because it quietly sets the ceiling on what you should be willing to spend to prevent the outage in the first place.

There is no honest industry average we can hand you, and you should be suspicious of anyone who offers one. The cost of an hour offline depends entirely on your business — a busy restaurant at Friday dinner and a two-person bookkeeping firm on a quiet Tuesday are not remotely comparable. What we can give you is a method to estimate your own figure, and a clear view of which prevention is cheap enough to be an easy decision.

Downtime is not one cost — it is five

The mistake is thinking of an outage as a single line item. It is at least five, and they stack on top of each other.

Lost sales. If customers buy from you in real time — a shop, a café, an online store, a clinic that bills per visit — an hour offline can mean an hour of revenue that never arrives. Some of it comes back later; a customer who could not check out at noon might return at two. Much of it does not. The person who found your checkout broken bought elsewhere and is not coming back today.

Idle payroll. Your staff are paid whether the systems work or not. When the tools they need are down, you are paying full wages for reduced or zero output. Five people at their desks unable to work is a direct, measurable cost for every minute of the outage, and it does not depend on whether you made any sales that hour.

Recovery labour. Someone has to fix it. That is time — yours, a staff member’s, or a provider’s — spent diagnosing, restoring, and confirming things work again. It rarely stops the instant service returns, either. There is usually a tail of re-entering data, reconciling what was missed, and answering the questions that piled up.

Missed deadlines and knock-on costs. Some outages land on the worst possible day: payroll runs, a tax filing, a client deliverable, a shipment that had to go out. The cost there is not the hour — it is the penalty, the lost contract, or the overtime you pay to catch up. These are irregular and easy to forget when you estimate, and they are frequently the largest number in the whole exercise.

Reputation and trust. The hardest to quantify and the easiest to dismiss. A customer who hits a dead website, a booking system that will not take their appointment, or a card machine that declines everything walks away with an impression. One outage is forgivable. A pattern teaches people to plan around you, and that erosion does not show up on any single day’s books.

How to estimate your own number

You do not need a spreadsheet full of assumptions. You need four figures you already roughly know.

1. Revenue per open hour. Take a normal month’s revenue, divide by the number of hours you are actually open and transacting. That is your baseline hourly exposure for anything that stops sales outright.
2. Loaded payroll per hour. Add up the wages of the people who would be sitting idle during an outage, per hour, including the employer costs on top of the headline wage. Not everyone is blocked by every outage — be honest about who actually is.
3. A recovery estimate. Guess how long a real incident takes to fix and who fixes it, then value that time. If you have never measured it, assume it is longer than you think, because untested recovery always is.
4. A deadline multiplier. Ask one question: what is the worst day of the month for this to happen, and what would it cost if it did? Keep that number separate. It is the one that turns a nuisance into a crisis.

Add the first three for a plausible ordinary hour, then hold the fourth as your worst-case. Now you have two numbers: what a typical outage costs, and what a badly timed one costs. Most owners find the gap between them is the real story — the average is survivable, the bad day is not.

One more honest adjustment: multiply by likely duration, not one hour. Outages are rarely tidy sixty-minute events. A failed drive, a ransomware hit, or a corrupted database can mean a day or more, and the cost scales with every hour you cannot bring things back. That is why the recovery figure above matters as much as the sales figure — the thing that ends the outage is a fast, proven restore.

The cheap prevention that avoids most of it

Here is the part that makes the exercise pay off. Once you have a number, the prevention that removes most of the risk turns out to be inexpensive relative to a single bad outage.

• Backups you have watched restore. Not backups that run — backups you have proven come back. The difference is the whole game, and testing a restore takes about an hour a quarter. We wrote the full method in why a backup is not a backup until you have watched it restore.
• Managed devices and patching. A large share of outages are self-inflicted: an unpatched machine, a failed update pushed at the wrong time, a laptop nobody was watching until it died. Keeping devices enrolled, patched, and monitored catches most of these before they become an outage. That is the core of what device management buys you.
• Redundancy on the things that stop revenue. A second internet connection, a spare card terminal, a documented manual fallback for taking orders when the system is down. None of these are expensive next to an hour of lost Friday-night sales.
• A written recovery plan with an owner. The cheapest prevention of all is knowing, in advance, who does what when a system goes down — so the outage does not also become an hour of confusion about who is responsible.

The through-line is that most downtime is preventable or shortenable, and the prevention costs a fraction of the exposure you just calculated. That is usually the moment a vague worry becomes a clear decision.

If you would rather not carry this yourself, this is exactly what a managed practice is for — the patching, the monitoring, the tested restores, all run and reported so an outage is short and boring instead of long and expensive. If you are weighing whether that is worth it, our note on what managed IT actually costs lays out the real ranges without the sales gloss.

Send us two paragraphs about the systems your business cannot run without, and we will reply in writing within one business day.