Tuesday, 9 in the morning. The file server is not responding, the business management program will not start, email works only on phones. By 10 it is clear this is not “something a restart fixes”, but a serious problem. The question every owner asks at that moment — “what is this costing us?” — has an exact answer. It is just that almost nobody has calculated it in advance.
All sorts of statistics about the average cost of downtime circulate on the internet. You will not find any of them here, for a simple reason: other people's average says nothing about your company. An accounting firm in the middle of a filing deadline, a warehouse that loads goods daily and a design office lose completely different amounts from the same hour of standstill.
Instead, you will find the method: how to calculate your own figure in about half an hour, which costs hide outside the obvious calculation and how to use the result — because this figure, once known, changes the way you look at the IT budget.
The basic formula: two components
The direct cost of an hour of downtime has two parts: the money that goes out anyway and the money that stops coming in.
Part 1: wages paid for work made impossible
The calculation: take the company's total monthly payroll cost (contributions included), divide it by the number of working hours in the month and multiply by the proportion of the team that cannot work without the fallen systems.
A worked example — the numbers are chosen round, so you can follow the logic, not a statistic: a company of 30 people with a total payroll cost of 300,000 lei per month has, at roughly 168 working hours in the month, a cost of about 1,800 lei per hour. If, when the systems go down, two thirds of the team can no longer actually work, you are paying around 1,200 lei per hour for work that does not get produced. Over an 8-hour day: nearly 10,000 lei — wages alone, before anything else. Redo the calculation with your numbers; it takes five minutes and it is the first half of the answer.
Part 2: the revenue that stops being produced
The second half depends on how the company earns its money. The right question is not “how much do we invoice per hour on average”, but “what exactly does not happen while we are down?”: orders not taken, goods not shipped, work not delivered on time, clients calling with nobody answering. Part of this revenue is recovered after the restart — the order is taken on Wednesday instead of Tuesday — but part is never recovered at all: the client in a hurry bought from someone else. Estimate honestly the proportion between the two for your business; even a rough estimate is more useful than none.
The costs that do not show up in the first tally
- Recovery: the overtime after the incident, the manual re-entry of the lost day's data, checking what was saved and what was not. The day of standstill is usually paid for once more, in the following week.
- Deadlines with penalties: contracts with delay clauses, tenders with a submission cut-off — and the tax obligations with fixed deadlines; in a company issuing many invoices, a prolonged outage presses directly on their legal submission deadline.
- The cost of the intervention: the emergency call-out paid at emergency rates, possibly parts or equipment bought “for yesterday”.
- Trust: the client who caught the bad day does not send you an invoice, but silently adjusts their willingness to depend on you. It is the hardest cost to measure and the only one that grows with every repeated incident.
Not all hours cost the same
The calculation above gives an average — reality has sharper edges. The same system failure costs completely different amounts depending on the moment: on payday, in the last days before a tax deadline, at the peak of the season or on Monday morning first thing, when the whole company starts its engines at once. When you assess your risk, do not ask only “what does a day cost?”, but also “what is the most expensive day it could happen to us on?” — because Murphy's law has a solid track record in IT.
How to cut the bill: two levers
The calculated figure is attacked from two directions: you make the outages rarer and you make them shorter.
Rarer: prevention
Monitoring that catches the full disk and the overheating server before the actual failure, updates done on time, equipment replaced as planned at end of life, not on the day it dies. It is also worth saying that architecture matters: where the servers sit and how they are duplicated radically changes the risk profile — we have separately compared the five-year costs of the cloud and on-premises server options, and availability is an important part of that comparison.
Shorter: recovery time
Here come two questions worth putting down in black and white: how quickly must we be running again (within hours? by the next day?) and how much data can we afford to lose (the last hour? the last day?). The answers dictate the technical solution: tested backup with a timed restore, spare equipment for the critical points, redundancy where the numbers justify it.
And a third question, often forgotten: who intervenes and within what time? If the answer is “we call someone and hope”, the recovery time is undefined by construction. An IT outsourcing contract with a committed response time turns the unknown into a parameter — and when assessing such a contract, the clauses that matter are exactly the ones about times and coverage; the frequently asked questions about prices and contracts show what to look at before you sign.
Use the figure: the budget conversation changes
Until now, the conversation about IT spending probably sounded like this: “do we really need all of this?”. With your figure on the table, it sounds different: you compare the annual cost of prevention with the cost of the downtime days you avoid. The calculation is yours, with your numbers — but its structure is always the same: if a day of standstill costs you as much as several months of preventive maintenance, the right question is no longer whether you can afford prevention, but how many days of standstill you can afford without it.
The same calculation also tells you where not to spend: if a system can sit idle for a week without hurting, it does not need expensive redundancy. The figure is not an argument for “more IT” — it is an argument for IT matched to the real risk.
The next step: do the calculation of the two components for your company today — wages per hour of standstill and unrealised revenue — and write the result somewhere visible. It is the figure that will make the decisions instead of emotions at the next budget discussion. Neoxis, an IT services company founded in 2015 in Pitești, works with SMEs across Romania, with remote support nationwide and on-site interventions in the Pitești–Argeș area.