What downtime costs depends on who you are. Published surveys of IT and business leaders put the average at roughly $14,000 to $31,000 per minute, depending on the survey and who was asked. For a small or mid-size website, we found no study that measures it. Your number is your own revenue per minute, plus staff time, plus a few direct costs, multiplied by how long you were down.

This post covers what five widely quoted studies actually measured, what the largest recent outage shows about cost, and how to calculate the cost of an outage on your own site.

What the major studies measured

Five widely quoted studies are behind many of the downtime-cost figures you will see. They ask different people about different things, so the numbers are not interchangeable.

StudyWho answered, and what was measuredHeadline result
Splunk and Oxford Economics, The Hidden Costs of Downtime, May 2026 2,000 executives at Global 2000 companies in 20 countries. Unplanned downtime, which the 2024 edition defined as “any service degradation or outage of a business system”3. $600 billion a year across the Global 2000, about $300 million per company2, up 50% from $400 billion in 2024. Average $15,000 per minute. Average share price drop of 3.4% after an incident. $95 million a year in lost revenue per organization1.
New Relic, 2026 Observability Forecast, fielded April and May 2026 2,575 engineering and IT leaders and practitioners in 24 countries and 12 industries. High-impact IT outages. Mean $1.85 million per hour ($30,833 per minute). An annualized $74 million for the businesses surveyed4.
Uptime Institute, Annual Outage Analysis 2026, May 2026 Respondents to Uptime’s 2025 Annual Survey. The analysis covers IT and data center outages. The cost of each respondent’s most recent major outage. 57% said more than $100,000. For the second consecutive year, one in five said more than $1 million5.
ITIC, 2024 Hourly Cost of Downtime Survey, November 2023 to March 2024 An independent web survey of over 1,000 firms worldwide, as ITIC describes it. Respondents’ own estimates of the cost of one hour of downtime, excluding litigation and civil or criminal penalties. Over 90% of mid-size and large enterprises put an hour above $300,000. 41% of enterprises put it between $1 million and over $5 million6.
EMA Research for BigPanda, IT outages: 2024 costs and containment More than 400 IT professionals in North America, EMEA and APAC. Unplanned IT downtime. Commissioned by BigPanda. Average $14,056 per minute, and $23,750 per minute for large enterprises9.

Splunk, New Relic and BigPanda sell observability or IT operations software. That does not make their numbers wrong, but you should know who is asking the question.

Why these numbers do not transfer to your website

  • The companies are much larger. Splunk’s $600 billion works out to about $300 million per Global 2000 company per year. ITIC’s headline figure is for mid-size and large enterprises. A typical website owner is not in either population.
  • None of the studies isolates a website. Splunk’s 2024 definition counts any outage or degradation of a business system. In its 2026 data, 43% of downtime events were network or IT environment related, 32% were cybersecurity-driven and 24% were application or infrastructure failures2. New Relic counts high-impact IT outages. Uptime counts IT and data center outages.
  • They are estimates, often rough ones. ITIC says its figures are the respondents’ estimates, and that only 22% of organizations, about one in five, can accurately assess the hourly cost of downtime7.
  • They include costs a small site rarely has. Splunk’s 2026 study also reports regulatory fines averaging $51 million per organization, ransomware payouts averaging $40 million, and share price drops1. These attach to large and listed companies.
  • Means are pulled up by large events. New Relic and EMA report averages, and a few very costly outages raise an average well above what a typical outage costs. EMA describes its goal as an average that is “defensible, not definitive”8.

One figure to ignore: you may still see $5,600 per minute quoted. EMA says that number came from a casual remark in a 2014 analyst blog and was never research8.

What the largest recent outage shows

On July 19, 2024, a CrowdStrike software update affected an estimated 8.5 million Windows devices, under 1% of all Windows machines, according to Microsoft11. Parametrix, a cloud monitoring and insurance firm, estimated the direct financial loss for US Fortune 500 companies (excluding Microsoft) at $5.4 billion. A quarter of the Fortune 500, 125 companies, were affected. The weighted average loss was $44 million per company, and it ranged from $6 million in manufacturing to $143 million for airlines. Parametrix estimated that cyber insurance would cover no more than 10% to 20% of the loss, or $0.54 billion to $1.08 billion10.

Delta, one of the airlines hit, put numbers on its own outage in an August 2024 SEC filing: about 7,000 flight cancellations over five days, an estimated $380 million direct revenue impact and $170 million of non-fuel expense for the September quarter, offset by $50 million lower fuel expense. It said it was pursuing claims for damages of at least $500 million12.

Worldwide totals of $10 billion and more circulate as estimates, but we found no primary source that measured one. Parametrix’s $5.4 billion is a modeled estimate for US Fortune 500 companies only.

The useful lesson is the spread. The same outage cost companies in one sector $6 million on average and companies in another $143 million, because the cost depends on how much of the business runs through the system that failed. That is why your own estimate beats any benchmark.

How to estimate what downtime costs your site

Work out the cost of one outage first, then multiply by how often it happens. For one outage:

Cost = D × R × (1 - B)  +  P × W  +  X
InputWhere to get it
D, minutes of downtimeYour incident log: the time from first failure to full recovery. That is time to detect plus time to fix. New Relic’s respondents averaged 41 and 54 minutes for high-impact outages; use your own history. Count a partial outage, such as a broken checkout, for the share of revenue that flows through the broken part.
R, revenue per minuteYour analytics for the same weekday and hour as the outage. Annual revenue divided by 525,600 minutes is an average that hides peaks and quiet hours. For leads or sign-ups, use the value of a conversion times conversions per minute.
B, share of lost sales that come back laterCompare orders in the 24 hours after a past outage with a normal day, as a number from 0 to 1. Use 0 if you do not know.
P and W, staff hours and their hourly costP is people times hours each: staff who could not work, plus staff who worked the incident. W is their pay plus overhead per hour.
X, direct extrasSLA credits and refunds you pay, ad spend sent to pages that were failing, and extra support or overtime costs.

A worked example with made-up inputs, not benchmarks. An online store takes $30 a minute in revenue at 7 p.m. on a weekday. A checkout failure goes unnoticed for 30 minutes and takes 60 more to fix, so D is 90. From past outages, about 20% of lost sales come back later (B = 0.2). Three people spend an hour each fixing it, so P is 3, at $60 an hour (W = 60). The store was paying $12 a minute for ads that landed on the failing page.

ItemCalculationAmount
Lost sales90 min × $30 × (1 - 0.2)$2,160
Staff time3 hours × $60$180
Wasted ad spend90 min × $12$1,080
Total$3,420 (about $38 per minute)

Churn and reputation are real but hard to price. In Splunk’s 2026 survey, 81% of technology leaders cited loss of customers as a consequence of downtime1. For your own estimate, track cancellations and support tickets in the weeks after an outage and add them as a separate line, rather than folding a guess into the total.

Our free downtime cost calculator does the first two parts with simpler inputs: annual revenue spread evenly across the year, plus idle wages. It leaves indirect costs out, so treat its result as a floor.

Then multiply by frequency: annual cost is outages per year times the average cost of an outage. If you promise 99.9% uptime, that allows 43.2 minutes of downtime in a 30-day month (see how to calculate downtime), so you can price a breach of that promise with the same formula.

What to do about it

  • Shorten the time to detect. In New Relic’s data the average high-impact outage took 41 minutes to detect and 54 to resolve, and 42% of organizations learn about disruptions through manual checks or customer complaints4. In Splunk’s survey, 47% of technology leaders said customers are often or very often the first to detect an outage or degradation1. An external check can only notice a failure when it next runs, so the check interval caps detection time, plus any delay you configure before alerting.
  • Remove repeat causes. New Relic’s respondents named third-party and cloud provider failures as the top cause of outages, followed by network failures and software change deployments4. Uptime reports that failure to follow established procedures remains the leading driver of human-error outages5. List the third parties your site depends on (payments, DNS, CDN, hosting) and put changes through a checklist with a rollback step.
  • Give customers somewhere to look. In Splunk’s survey, 90% of technology leaders reported more demand for customer support after downtime1. A status page gives customers a place to check before they contact you.
  • Write your cost per minute down. Once you have the number, you can compare it with what monitoring, on-call coverage or redundancy cost, instead of arguing from a benchmark built on a different kind of company.

For incident counts from 41 cloud, AI and SaaS providers, see our Q3 2026 Outage Report.

Find out about downtime before your customers do

Site Qwality runs HTTP, TCP port, ping, TLS and DNS checks on the schedule you set and alerts you by email, SMS, Slack and other channels when one fails. The free plan includes 100,000 checks a month at 5-minute intervals from one location, plus one status page. Paid plans start at $15 a month for 500,000 checks at 30-second intervals, from as many locations as you choose. No credit card to start.

Start free

Sources

  1. The $600 Billion Wake-up Call: New Splunk Research Reveals Downtime is a Systemic Business Crisis, Cisco Newsroom, May 19, 2026
  2. The Hidden Costs of Downtime 2026: A $600 Billion Wake-Up Call, Splunk
  3. .conf24: Splunk Report Shows Downtime Costs Global 2000 Companies $400B Annually, Splunk, June 11, 2024
  4. New Relic Report Finds High-Impact IT Outages Drain $74 Million Annually from Businesses, New Relic, September 22, 2026
  5. Uptime Announces Annual Outage Analysis Report 2026, Uptime Institute, May 13, 2026
  6. ITIC 2024 Hourly Cost of Downtime Report Part 1, ITIC, September 3, 2024
  7. ITIC 2024 Hourly Cost of Downtime Part 2, ITIC
  8. IT outages: 2024 costs and containment, Enterprise Management Associates (EMA)
  9. The rising costs of IT downtime, BigPanda (summary of the EMA research)
  10. CrowdStrike to Cost Fortune 500 $5.4b; Insured Loss Range of $0.54b - $1.08b, Parametrix, July 24, 2024
  11. Helping our customers through the CrowdStrike outage, The Official Microsoft Blog, July 20, 2024
  12. Delta Air Lines Form 8-K, Item 7.01, U.S. Securities and Exchange Commission, August 8, 2024