"Our site was down 47 minutes, but that's not critical, right?" Wrong. Those 47 minutes cost real money — lost revenue, burned engineering hours, customers who left for a competitor and won't return. The problem is most teams don't know the real cost of their downtime, so they can't justify investments in reliability.
Below: concrete numbers for the cost of downtime across different business types, a formula to calculate your specific cost, hidden expenses that 90% of companies miss, and what to do to minimize losses.
The cost per minute of downtime: real data
Downtime cost depends on three factors: business size, business type, and time of day (peak hours vs. night). Here's data from industry research and public reports:
| Segment | Cost / minute | Cost / hour | Source |
|---|---|---|---|
| SMB (up to 100 employees) | $427 | $25,620 | Gartner |
| Mid-market SaaS | $5,600 | $336,000 | ITIC Survey |
| E-commerce (top 100) | $11,000 | $660,000 | Statista |
| Financial services | $16,000 | $960,000 | Ponemon Institute |
| Amazon (public data) | $220,000 | $13.2M | Amazon SEC filings |
Important: these are averages. Your actual cost depends on the specific moment. Downtime on Friday evening for e-commerce costs 3-5x more than Monday morning. Downtime during a promotional campaign costs even more, because you lose not just sales but marketing investments.
Formula for calculating downtime cost
To calculate the cost of downtime for your business, use this formula:
Total Cost = Direct Loss + Labor Cost + Hidden Costs
Direct Loss = (Hourly Revenue / 60) x Minutes Down
- E-commerce: lost transactions x average order value
- SaaS: at-risk MRR x churn probability (typically 5-15%)
- Ad-supported: lost pageviews x RPM
Labor Cost = Engineers on incident x hourly rate x time spent
- Include: detection time + response time + fix time + post-mortem
- Typical: 3-5 engineers x 2-6 hours = 6-30 engineering hours
Hidden Costs = Direct Loss x 1.5-2.0 multiplier
- Customer support surge
- SLA credit payouts
- SEO ranking impact
- Customer lifetime value loss from churn
- Delayed roadmap (engineers fixing instead of building)
Example: SaaS service with $100K MRR
Incident: API unavailable 45 minutes during business hours (Europe)
Direct loss: ($100,000 / 30 / 24 / 60) x 45 = $104 (direct MRR loss is negligible for SaaS)
Churn risk: 200 affected users x 10% churn probability x $50 avg MRR = $1,000/month recurring loss
Labor: 4 engineers x 4 hours x $80/hr = $1,280
SLA credits: $2,000 (10% monthly for affected enterprise clients)
Support: 150 tickets x 15 min avg = 37.5 hours x $30/hr = $1,125
Total: ~$5,500 one-time + $1,000/month recurring churn = $17,500 first year
Hidden costs everyone forgets
SEO damage. Google demotes sites with frequent 5xx errors. One major outage can drop rankings for 2-4 weeks. For a site with 10,000 organic visits/month, that's a loss of $3,000-10,000 in organic traffic. More: causes and prevention of downtime.
Opportunity cost. Every hour spent firefighting is an hour not spent building new features. If your 5-engineer team spends 20% of time on incident response instead of development, you lose the equivalent of one full-time engineer ($120-200K/year).
Reputation. 37% of users say they won't return to a site after a bad experience (Google/SOASTA research). For B2B SaaS this is even more critical: enterprise customers evaluate reliability when choosing vendors, and one public outage can cost a $100K+ contract.
Morale. Teams that regularly fight fires at night burn out. Turnover in "fire-fighting" teams is 40% higher than average (DevOps Institute). The cost to replace a senior engineer is 6-12 months of salary.
How to minimize downtime cost
You can't guarantee zero downtime — any system eventually breaks. But you can radically reduce three multipliers in the formula: incident frequency, detection time, and recovery time.
Reduce detection time (MTTD). Without monitoring you learn about problems from users — 15-45 minutes later. With multi-region monitoring — 30-60 seconds. For a business with $5,000/hour downtime cost, that's $1,200-$3,700 saved per incident.
Prevent preventable outages. SSL monitoring catches expiring certificates 30 days ahead. DNS monitoring catches accidental record deletions in minutes. Cron monitoring catches stuck backup jobs before data loss.
Eliminate false alarms. Every false positive costs engineer time to check "is something actually broken?". At 5% false positive rate and 10 alerts/week, that's 0.5 engineer-days wasted. Quorum confirmation reduces false positives to less than 0.1%.
Communicate fast. A status page with real-time updates cuts support load by 60-80%. Customers see "we know, we're working" instead of "try reloading".
ROI of monitoring: the numbers
Monitoring isn't an expense — it's insurance with measurable ROI. Here's the math:
Scenario: SaaS with $5,000/hour downtime cost, 3 incidents/year, MTTD without monitoring — 30 min.
Without monitoring: 3 incidents x 30 min faster detection = 0 savings (baseline)
With monitoring (MTTD 1 min): 3 incidents x 29 min saved x $83/min = $7,221 saved/year
+ Prevented incidents (SSL expiry, DNS, cron): ~2 prevented/year x $5,000 = $10,000 saved
+ Reduced false alarms: 50 hours/year saved = $5,000 saved
Total savings: ~$22,000/year
Cost of monitoring: $60-600/year → ROI: 36x-366x
Downtime will happen. The question is: how quickly do you detect it, how quickly do you recover, and how much do you lose along the way. AtomPing's free plan covers 50 monitors with multi-region checks, quorum detection, and status pages — enough to reduce MTTD from 30 minutes to 30 seconds. Use the uptime calculator to determine your acceptable downtime for your SLA.
FAQ
How much does downtime cost per minute?
It varies dramatically by business size. For SMBs, the average is $427-$1,000 per minute. For mid-market companies, $5,000-$16,000 per minute. For enterprises like Amazon or major banks, losses can exceed $100,000 per minute. These figures include lost revenue, recovery labor, customer churn, and reputational damage — not just the direct sales impact.
What are the hidden costs of downtime?
The direct revenue loss is often less than half the total cost. Hidden costs include: engineering hours spent on incident response and post-mortems, customer support surge (ticket volume spikes 5-10x), SLA credit payouts, SEO ranking drops (Google penalizes sites with frequent downtime), long-term customer churn (customers who experience outages are 3x more likely to switch providers), and opportunity cost of delayed feature work.
How do I calculate downtime cost for my business?
Use this formula: (Revenue per hour / 60) + (Affected employees x Average salary per minute) + (Estimated reputation cost per minute). For e-commerce: add lost transactions x average order value. For SaaS: add at-risk monthly recurring revenue x churn probability. For ad-supported: add lost ad impressions x CPM. Always multiply by 1.5-2x for hidden costs.
Does downtime affect SEO rankings?
Yes. Google's crawlers check your site regularly. If they encounter 5xx errors or timeouts, your crawl budget decreases and rankings can drop within days. Extended outages (4+ hours) can cause pages to be temporarily de-indexed. Recovery takes weeks — much longer than the outage itself. Regular short outages are also harmful, as they signal unreliability to Google's quality algorithms.
How much downtime is acceptable?
It depends on your SLA. 99.9% uptime allows 43 minutes of downtime per month — that's considered the baseline for production services. 99.5% allows 3.65 hours. The acceptable amount depends on your business: an e-commerce site during Black Friday has zero tolerance, while an internal dashboard can tolerate more. Use an uptime calculator to convert percentages to real hours.
What's the ROI of uptime monitoring?
If monitoring detects an outage 15 minutes faster than user reports, and your downtime costs $5,000/hour, that's $1,250 saved per incident. Most production services experience 2-4 significant incidents per year, so monitoring saves $2,500-$5,000 annually — far more than the cost of any monitoring tool. Factor in prevented outages (SSL expiry alerts, DNS change detection) and the ROI multiplies further.