SME owners often ask: how much is it worth spending on backup, a failover link or server redundancy? The answer starts with a completely different question — how much does an hour cost us when a critical system is down? Most companies have never actually calculated it.
In boardroom conversations we usually hear one of two extremes. The first: "we survived the last outage, nothing dramatic". The second: "we can't afford any downtime at all, everything has to be in HA". Both are worth almost nothing until a concrete number sits underneath. Without it, any discussion of reliability is ideology, not a business decision — and the kind of ideology that generates unnecessary spending on one side and unacceptable risk on the other.
The good news: you don't need a consultant or a Monte Carlo model. Calculating the cost of an hour of downtime with reasonable accuracy takes one afternoon and three internal conversations.
What actually gets lost during downtime
An IT outage in an SME generates costs in four pockets at the same time. Most companies only count the first — which is why their number comes out unrealistically small.
- Lost productivity. People sit there but accomplish nothing useful. We count gross salaries plus overheads, multiplied by the share of people genuinely blocked and by their productive portion of a normal day.
- Lost or deferred sales. Orders not taken, quotes not sent on time, a customer who called and didn't get through — some of this returns after restart, some drifts to competitors. Not revenue, but margin.
- Recovery costs. Overtime for the team catching up, an emergency call-out for the admin or external vendor at weekend rates, courier shipments for things that should have gone with regular delivery.
- Contract penalties and trust erosion. SLAs with a chain retailer, late-delivery fines, a client who already heard "the system is down" once and will compare prices with competitors next time. This last one is the hardest to quantify and the most painful long-term.
Importantly — these four pockets behave differently over time. Productivity is linear (the second hour costs about the same as the first). Sales has a threshold (the customer waits for 30 minutes, calls a competitor after two hours). Recovery costs rise in steps (Monday morning and Friday evening are two different worlds). Contract penalties typically kick in at daily boundaries.
Calculating it with a concrete example
Take a realistic case we saw at one of our clients. A 30-person electrical wholesaler near Łódź, Comarch Optima as the ERP, a simple WMS integrated with Optima. One Monday at 11:00 the database server dies — a disk failure. The ERP is back at 15:00. Four hours of downtime at peak time.
Four pockets:
- Productivity. Of 30 people, 22 work with the system (sales, back office, logistics). Average salary plus overheads ~85 PLN/h. Not fully unproductive — some stay on the phone, some tidy paperwork — assume 60% loss. 22 × 85 × 4 × 0.6 = 4490 PLN.
- Sales. The business does ~180,000 PLN of daily revenue at ~18% margin. Four peak hours is around 90,000 PLN of revenue. After restart, 40% of orders return (customers who needed to order today), 60% drift away or go to competitors. Lost margin: 90,000 × 0.6 × 0.18 = 9720 PLN.
- Recovery costs. External admin on emergency call-out — 6 hours at 250 PLN = 1500 PLN. Tuesday-evening overtime catching up — 10 people × 2 h × 70 PLN with overtime loading = 1400 PLN. Total 2900 PLN.
- Penalties and trust. One shipment to a chain retailer with an SLA code — 2500 PLN penalty for missing the delivery window. One steady customer cancelled a ~15,000 PLN order and moved to a competitor over the following months; conservatively 2000 PLN of lost margin per year for three years. For this specific outage: 4500 PLN.
Total: around 21,600 PLN for four hours, averaging 5400 PLN per hour of downtime at peak time. That's a number you can work with. Importantly — without the fourth pocket you'd get 4200 PLN per hour and investment decisions would come out differently.
Not every system deserves the same protection
The biggest mistake after getting this number is applying it to everything. "An hour of ERP costs 5400 PLN, so an hour of HR must too." No.
Make a simple matrix. First column — system (ERP, email, phone system, website, HR system, warehouse management, file server). Second — how many people and processes genuinely stop when that system is down. Third — how quickly the customer notices. Fourth — the per-hour cost calculated as above.
In the same 30-person SME, three tiers typically emerge:
- Critical (ERP, WMS, phone system for sales): 3,000–6,000 PLN per hour. RTO in hours, max 4–8 hours during business days.
- Important (email, website with contact form, HR system during payroll week): 500–1,500 PLN per hour. RTO: half a day to a day.
- Useful (intranet, wiki CMS, holiday request system): realistically below 500 PLN per hour. RTO: 2–3 business days without drama.
Only now can you consciously decide where to invest in HA, where in a backup with fast RTO, and where a daily backup is enough. A company that treats all three tiers the same either overpays for protecting the holiday system or under-protects the ERP. Usually both at once.
How to use this number in an investment decision
The same 30-person wholesaler faced a decision: keep a single server with 24-hour backup, or invest in a second server in HA configuration (one-off ~35,000 PLN plus ~6,000 PLN per year). Without the number, one side was "we're not spending 40k, that's expensive", the other "HA is obviously better". With the number, the conversation changes.
Historically, the company had an average of two serious ERP outages per year, averaging 3–4 hours each. That's about 7 hours × 5400 PLN = ~38,000 PLN of losses per year. The HA investment pays back in a year in a no-improvement scenario, and if it avoids one big weekend outage (10+ hours) over three years, it pays back twice. Decision: invest, but only in ERP and WMS, not in the other systems. For email, a cloud service with a 99.9% SLA plus backup is enough. For the intranet — a daily backup.
The same logic cleanly explains the opposite decisions too. If the per-hour cost of the corporate website (no e-commerce, just a business card) works out to 200 PLN, there's no point buying a 20k-PLN cluster. Decent hosting with monitoring and backup is fine.
What next
If you've never calculated the hourly cost of your ERP going down — don't start with a 40-row Excel model. Do this first calculation on a sheet of A4 paper, in four pockets, at a table with your finance person and someone from the department that depends on the system most. It takes two hours. It settles a decision you've probably been walking around for months.
If you'd like someone to run this calculation at your company without selling you a specific solution at the same time — that's one of the things we're here for. The first call is free and takes half an hour.
Without a number for the cost of an hour of downtime, every conversation about IT reliability is ideology, not a business decision.