What does a 24-hour outage cost?

Linear Versus Realistic

24-hour outage cost by company size

The table presents two columns. The linear column is the per-hour benchmark times 24. The realistic column adjusts downward for the fact that off-business-hours traffic is lower and productivity loss flattens once people have gone home. Both are useful: the linear figure is the defensible upper bound for the calculation, the realistic figure is the number you would actually expect to see in a P&L impact reconciliation.

Realistic figures discount the off-hours block by roughly 35% but add a reputation premium of 15 to 20%. Final ratio versus linear is 0.75 to 0.85x for most segments. Sources: ITIC 2024, our own modelling. Figures USD, as of 2026-05-18.

Historical Reference Events

What 24-hour-class outages actually looked like

Truly 24-hour-and-up outages of major branded services are rare enough to enumerate. The table below lists every major outage of the past 15 years that either crossed 24 hours of full impact or had a recovery tail long enough to qualify for full-day cost analysis. Note that several of these are technically shorter outages with longer recovery tails, which is the more typical pattern.

The Recovery Tail Problem

A 24-hour outage is rarely 24 hours of one event

Most 24-hour outage cost figures are not the cost of 24 continuous hours of total downtime. They are the cost of an initial severe outage (often 2 to 8 hours) followed by a long recovery tail in which the service is degraded, partially restored, then re-degraded, with intermittent capability for the rest of the day. This shape is operationally distinct from a single 24-hour blackout.

The AWS us-east-1 outage of 7 December 2021 illustrates this. The acute event was approximately 7 hours of severe API degradation. Some downstream customers, particularly those whose own architectures had cascading single-AZ dependencies inside us-east-1, did not see full recovery for more than 30 hours after the initial incident started. For those customers the cost calculation is appropriately modelled as a 24-hour-class event even though AWS's own post-mortem describes a shorter root-cause window.

For DR planning, this means the 24-hour case is not just "everything was off for a day". It is "the service was unreliable for a day, with periods of partial restoration that complicated incident communications, made user retry behaviour expensive, and produced data-consistency tail problems for hours after the underlying systems were nominally healthy".

Customer Churn Tail

The line that hits in Q+1

Customer churn from a public 24-hour outage typically runs 5 to 10% incremental over the 90 days following the incident in B2C. For a $100 ARPU subscription business with 1 million subs, 5% incremental churn over 90 days is roughly $1.25M of annualised recurring revenue lost. That number lands in the following quarter's board pack as "higher than expected churn", often without being correctly attributed to the outage that caused it.

B2B SaaS sees a smaller headline churn (1 to 3% incremental), but a meaningful renewal pricing concession averaging 5 to 15% off the next contract value as customers use the incident as leverage. For a customer book worth $50M ARR, a 10% concession on the 30% of customers up for renewal in the next 12 months is $1.5M of recurring revenue erosion. This line is usually invisible in incident post-mortems because it does not show up until the next renewal cycle.

For the methodology behind these estimates, see how we calculate downtime cost. The churn coefficient is the most contested component of any downtime cost model, and we treat it as a hedged estimate rather than a fixed figure.

DR Planning

How to use the 24-hour figure in your DR business case

For a DR investment business case, the 24-hour scenario is the case that makes the math work. The shorter scenarios (1 hour, 4 hours, 8 hours) often do not justify the cost of multi-region active-active because the expected annual loss is comparable to the annual cost of the redundancy. The 24-hour scenario reverses this. Even with a low annual probability (say 5%), the expected annual loss of $360,000 to $1.2M for a mid-size enterprise comfortably exceeds the $100,000 to $250,000 annual cost of meaningful multi-region redundancy.

The right framing for the board is: "In any given year we have a 5% chance of a 24-hour outage. Expected loss is $360,000. The investment to push the probability below 1% costs $200,000 annually. Net expected benefit is $360,000 minus $200,000 minus residual loss, or roughly $90,000 per year in expected value plus the option value of avoiding a catastrophic single-event impact." Use the business case builder to walk this for your inputs.

Frequently Asked

Common Questions