Cost Of
Cost of downtime
An uptime percentage is hours of outage in disguise. Multiply by revenue per hour, add the people on the incident.
What downtime costs per hour and per year
An uptime target is a downtime budget written in a friendlier font. 99.9% sounds like a rounding error and allows 8.8 hours of outage a year; 99% allows 3.6 days. The bill for those hours is lost revenue plus the fully-loaded cost of everyone pulled into the incident.
This calculator turns an uptime percentage into hours, then into money, and shows what one hour down costs and what the next nine on the target would save. It assumes a 24/7 service with outages spread evenly, no SLA credits and no churn — which makes it a floor, not a ceiling.
How the number is built
- Hours down
- 8,760 hours × (1 − uptime%). 99.9% is 8.8 hours a year, about 44 minutes a month.
- Revenue at risk
- annual revenue ÷ 8,760 × the share of revenue that stops when you are down (100% for a store or a metered API; less if orders queue and complete later).
- Incident cost
- hours down × people on every incident × fully-loaded hourly cost.
- Also shown
- the cost of one hour down, what reaching the next uptime tier would keep, and what halving time-to-recover would keep.
When to use it
-
Setting an SLA you can afford
Each additional nine costs engineering time. Pricing the outage hours it removes tells you which nine is worth buying.
-
Prioritising reliability work
Runbooks, alert routing and faster rollbacks all shorten outages. Halve the recovery time in the calculator and the saved amount is the value of that work.
-
Justifying on-call compensation
The incident line prices the people, not just the revenue. It is a concrete argument for paying for on-call properly, or for reducing how many people each incident pulls in.
Questions
How much downtime does 99.9% uptime allow?
8.76 hours a year — about 44 minutes a month. 99% allows 3.65 days a year, 99.95% allows 4.4 hours, and 99.99% allows 53 minutes. The calculator converts any percentage between 90% and 99.999%.
How do I estimate revenue lost per hour?
Start with annual revenue divided by 8,760 hours, then set the share of revenue that actually stops during an outage. A metered API or an online store is close to 100%; a business where orders queue and complete later is much lower.
Should I use average or peak hours?
The calculation spreads revenue evenly across the year, which understates an outage during peak traffic and overstates one at 4 a.m. If your outages cluster in business hours, raise the revenue-at-risk share to compensate.
Does it include SLA credits, churn and reputation?
No. Those are usually the expensive parts of a serious outage, and they are too company-specific to guess. Treat the total as the floor of what an outage costs.
Why count the responders?
Because incidents consume the most expensive hours in the company, usually out of hours, and the cost repeats every time. Three people on every incident, all year, is a line worth seeing next to the revenue number.