How much does an outage cost?

Introduction: an existential threat

In July 2024, cybersecurity firm CrowdStrike released a software update that inadvertently triggered widespread system crashes, causing software outages in apparently unconnected systems around the world. Airlines, banks, and medical services were disrupted for hours, with knock-on effects lasting days or weeks. The financial impact was both swift and staggering: Delta Airlines alone reported $380 million in revenue loss from canceled flights and reimbursements during the five-day incident. CrowdStrike’s stock plummeted over 11% that day, a stark market reaction to a single critical logic error caused by a faulty update.

While the company began its recovery quickly, the episode underscored a harsh reality of today’s digital world: that a single software failure can cascade across industries and inflict massive, often irreparable, damage.

And CrowdStrike isn’t alone. Over the past decade, outages have brought operations to a standstill at some of the world’s most recognized companies. Facebook’s 14-hour platform blackout in 2019 disrupted billions of users and cost an estimated $90 million. In December 2022, Southwest Airlines grounded over 16,700 flights during a weeklong operational meltdown, leading to a fourth-quarter net loss of $226 million and a total estimated cost of $800 million.

"We are deeply sorry", said Southwest CEO, Bob Jordan. "We have swiftly taken steps to bolster our operational resilience... and are re-examining the priority of technology and other investments planned in 2023."

Unfortunately for Bob, outages like these are more than just embarrassing; they’re financially devastating. And yet, despite the rising frequency and cost of such failures, more than 90% of companies either don’t know, or can’t calculate, the true cost of their most recent outage, let alone prepare for the next one.

So, how much could one failure really cost your company? Or, a better question, perhaps: can you afford not to know?

How much downtime can you afford?

Today, over 90% of organizations require 99.99% availability for their critical systems, according to ITIC. This standard, known as “four nines” reliability, permits no more than 52.6 minutes of unplanned downtime per year, or just 4.33 minutes per month.

The table below illustrates how various availability thresholds translate into actual downtime allowances. At 99.0% availability, for example, businesses allow 3.5 days of downtime each year – a level generally unacceptable for mission-critical infrastructure. At 99.999%, there’s barely enough room for a 30-second hiccup per month.

But achieving “four nines” reliability is no small feat. Systems are more interconnected and complex than ever before, and the sources of failure are often unpredictable.

Put simply, the problem is that you just don’t know what you don’t know. No matter the caliber of your engineers, mistakes happen. And the nature of applications today – distributed, complex, highly-interdependent – means the trigger may be entirely out of your control: cyberattacks, service failures, and bugs or misconfigurations in third-party systems are all common culprits. As the Uptime Institute notes in their Annual Outage Analysis for 2023, “system and software outages are partly caused by the complexity and scale of modern systems.” As system complexity grows, so does the number of unknown unknowns that could take your system down.

Estimating the true cost of failure

The cost of outages is hard to estimate, because we’re fundamentally trying to put a number on an absence. Companies are embarrassed about them, and reliable data is scarce.

At the same time, not putting a number on outages risks turning a blind eye to them and falling into the same trap again. And failing to quantify a risk doesn’t eliminate it. Instead, knowing what an outage costs tells you what you’re willing to spend to buy down the risk, and informs planning around mitigating it – whether by investing in reliability, buying insurance, or simply knowing that there should be an outage contingency in the company’s bank account.

We therefore propose a framework that companies should use to consider the risk they take on when they take actions that make an outage more likely. This framework consists of 3 categories of direct costs, and 3 categories of indirect costs. If your company has not at least investigated each of these categories, you are almost certainly underestimating the true cost of downtime.

Direct costs: the obvious financial hit

1. Lost revenue

Gartner estimates that outages cost businesses $5,600 per minute, or $330,000 per hour, in lost revenue alone.

For larger companies, the price tag is far steeper. According to ITIC, 40% of enterprises report that a single hour of downtime costs between $1 million and $5 million, excluding any legal costs or penalties. In key verticals such as finance, healthcare, transportation, and media, however, the average loss per that same 60 minutes can exceed a whopping $5 million, with ease.

2. Cost to fix

Outages take time to fix, and time that engineers spend on fixing outages is time they’re not spending on revenue-generating feature work. The toll on engineering time is real, but whether it looks more like a direct or an indirect cost depends on the company’s perspective.

One possible rule of thumb for how much it costs to fix an outage is simply the number of developer-days spent on a resolution multiplied by their prorated comp. A recent survey pegged the fully loaded cost of an FTE engineer at $300,000/year, or approximately $1,300/developer-day.

However, many companies view engineering salaries as a fixed, sunk cost. In this case, the cost of fixing that outage is the opportunity cost incurred by taking the engineers away from feature-work. One large survey recently estimated annual revenue per Engineer applied to feature work at $1.29 million/year – approximately $5,600/day (Source: DX Report Revenue-per-Engineer April 2025) – a significant figure that never appears on the actual bottom line.

3. Regulatory and legal costs

These losses are only part of the equation. Many outages also trigger compliance violations, with costly legal repercussions. According to industry research, the average organization loses $22 million each year due to fines and penalties related to system failures and compliance violations.

These costs are expected to climb as governments around the world introduce stricter regulations aimed at enforcing digital resilience. New legislation, like the EU’s Digital Operational Resilience Act (DORA), now requires financial institutions to demonstrate that their technology systems can withstand disruptions. Non-compliance can result in significant financial and reputational penalties, making regulatory risk one of the most pressing concerns in a post-outage environment.
In total, companies in the Global 2000 spend:

• $22M/year on regulatory fines
• $16M/year on SLA penalties
• $15M/year on settlements and legal costs

Indirect costs: the fallout

4. Reputation damage and lost trust

When systems go down, so does trust. Customers, partners, and investors start asking the same question: “If it happened once, will it happen again? And when?” That doubt can be more damaging than the downtime itself.

According to Oxford Economics, 40% of CMOs say that outages directly reduce customer lifetime value, meaning fewer repeat purchases, weaker brand loyalty, and lower overall revenue per customer. To contain the reputational fallout, Global 2000 companies spend $14 million per year on brand repair efforts, plus an additional $13 million on public relations and investor communication — a sign of how serious and expensive trust recovery can be.

For companies providing essential B2B services or infrastructure, relationships can be even harder to repair. Fastly experienced this firsthand after a major outage in 2021, with then CEO Joshua Bixby stating:

“We managed through a significant outage that impacted our Q2 results and will have an impact on our Q3 and full year outlook. We have a couple of customers, one of them being a top 10 customer, that have yet to return their traffic to the platform.”

Businesses need to embrace the reality that a single outage can do more than disrupt this quarter’s revenue. An outage can shift long-term customer relationships and stall product momentum. Even customers who were not directly impacted may reconsider their roadmap or vendor partnerships due to heightened concerns about reliability.

5. Organizational disruption

Damage to customer relationships is only the most visible fallout from outages, but the internal bleeding often does more long-term damage.

Every outage or critical system bug is an engineering incident – on-call pagers going off, war rooms convened, entire teams disrupted for days or weeks. This doesn’t just delay the product roadmap. The high-stress environment and context switching erode engineering productivity and morale, ultimately leading to turnover and attrition. Worse still, this damage is done even when a critical bug fails to cause an outage – because engineering still has to scramble to fix the issue in time.

The most lethal aspect of this dynamic is that the best engineers in an organization bear the brunt of this, because they’re the “fixers” who get called in no matter what breaks. These top-tier engineers tend to get parachuted into one stressful situation after another, and they’re the ones who are easiest to lose and hardest to replace.

The exact value of such disruptions is hard to calculate and highly context dependent, but here are some public indicators assembled by third parties:

• The cost of employee attrition and replacement has several components, per this framework from Dockyard (based in turn on a calculator from SHRM):
  • Cost of recruiting
  • Cost of time-to-fill
  • Cost of time-to-productivity
• Of these, the cost of recruiting is the easiest one to benchmark, but generally also the lowest – 20-25% of the engineer’s starting annual salary

The impact of a major outage is also felt beyond engineering. All customer facing teams shift from selling to damage control – incident messaging, communicating about blast radius and remediation timelines, and mending fences. If strategic customers are affected, this often takes a toll on executives’ time as well.

6. Cost of volatility to investors

Investor sentiment doesn’t come with a grace period. Downtime often prompts immediate and unforgiving reactions in the stock market – even when companies put together comprehensive remediation plans afterwards.

On average, public companies that experience unplanned downtime can expect to see their stock prices fall by 1% to 9%, with an average of 79 days needed for shares to recover to pre-incident levels (assuming they recover at all).

For shareholders, these events introduce volatility. For leadership, they erode investor confidence and can jeopardize funding, valuation, and strategic momentum.

Conclusion: downtime is a business risk – not a technical one

Outages can wipe out years of progress in minutes. From Southwest Airlines’s $800 million loss to CrowdStrike’s high-profile stumble, the examples are numerous – and growing. As digital infrastructure becomes more complex, companies must evolve their approach to reliability.

Time to detect, time to mitigate, and time to remediate are all critical indicators of operational maturity. And each one requires more than a reactive fix. Here are a few takeaways that stand out when it comes to building resilient systems:

• Testing isn’t optional – it’s existential.
• Incident response isn’t enough – prevention is ultimately key.
• The risk can’t be perfectly quantified – but not thinking about it is even worse.

Antithesis helps companies eliminate catastrophic software failures before they happen.

Our autonomous testing platform uncovers “unknown unknowns” in your code for you, finding and helping you fix the bugs other tools can’t…before they impact production.

Whether you’re running a database, trading platform, or an airline booking system, Antithesis gives you the confidence to ship faster and sleep better.

Let’s talk about what Antithesis can save you.

Appendix: Costs of major outages

Note that most of these estimates are for direct costs only, and do not account for indirect costs.

Other references

Numerical figures in this resource taken from:

• Rachel Dines (May 2011). State of Enterprise Disaster Recovery Preparedness [White Paper]. Forrester Analyst.
• Oxford Economics (2024). The Hidden Costs of Downtime: The $400B Problem Facing the Global 2000 [Consulting Report]. Oxford Economics.