BackupDrill
Start free

Backup and restore failure statistics (2026)

Most teams have backups. Fewer have checked that those backups restore, and the gap shows up on the worst day. These are the numbers on restore failures, restore testing, downtime costs and real incidents, each traced to the survey or report it came from.

75 statistics, each linked to its original source. Last verified October 17, 2026.

Key numbers

  • 94%

    94% of organizations hit by ransomware said the attackers tried to compromise their backups.

  • 57%

    57% of those attempts to compromise backups succeeded.

  • 66%

    66% of organizations whose data was encrypted in a ransomware attack used backups to recover it, up from 54% the year before (respondents could give more than one answer).

  • 42%

    42% of US organizations that experienced data loss were able to restore all of it.

  • 72%

    In a Q1 2025 survey of 285 German small and mid-sized businesses run by data recovery firm DATA REVERSE, 72% said they rarely or never test whether their backups can be restored. Only 28% test at least once a quarter.

  • 28%

    90% of organizations say they're confident they can recover from a cyber incident, but only 28% of ransomware victims whose operations or data were hit fully recovered all affected data.

  • 15%

    15% of IT professionals in Kaseya's 2025 survey test their backups daily, and another 25% test weekly.

  • 1 in 5

    One in five organizations reported that their most recent major outage cost more than $1 million.

  • None of 5

    When GitLab went to restore in 2017, none of its five backup and replication methods was working reliably.

  • 14 days

    Some Atlassian customers were without their products for up to 14 days while their deleted sites were restored. Atlassian had backups, but by its own account "the recovery process for restoring deleted sites at such a scale wasn't well defined."

Can backups actually be restored?

  • 66%

    66% of organizations whose data was encrypted in a ransomware attack used backups to recover it, up from 54% the year before (respondents could give more than one answer).

    Source: Sophos, “The State of Ransomware 2026”, September 2026

    Survey of 2,158 IT and security leaders whose organizations were hit by ransomware, run by Vanson Bourne for Sophos; 1,214 answered this question.

  • 6-year low

    In Sophos's 2025 survey, the share of victims recovering encrypted data from backups hit its lowest point in six years.

    Source: Sophos, “The State of Ransomware 2025”, June 24, 2025. Also: Sophos, “The State of Ransomware 2026”, September 2026

    Survey of 3,400 IT and security leaders in 17 countries; the blog post gives no percentage, and Sophos's 2026 report puts the 2025 figure at 54%.

  • 48%

    48% of organizations whose data was encrypted paid the ransom, and 2% got no data back at all.

    Source: Sophos, “The State of Ransomware 2026”, September 2026

    Some organizations both paid and restored from backups, so this can't be added to the share that used backups.

  • 94%

    94% of organizations hit by ransomware said the attackers tried to compromise their backups.

    Source: Sophos, “The impact of compromised backups on ransomware outcomes”, March 26, 2024

    Survey of 2,974 IT and security professionals, run by Vanson Bourne for Sophos in early 2024.

  • 57%

    57% of those attempts to compromise backups succeeded.

    Source: Sophos, “The impact of compromised backups on ransomware outcomes”, March 26, 2024

    This is the success rate among attempts, not the share of all organizations whose backups were compromised.

  • 85%

    85% of organizations whose backups were compromised also had their data encrypted, versus 52% of those whose backups were untouched.

    Source: Sophos, “The impact of compromised backups on ransomware outcomes”, March 26, 2024

  • 67%

    67% of organizations with compromised backups paid the ransom, compared with 36% of those whose backups were intact.

    Source: Sophos, “The impact of compromised backups on ransomware outcomes”, March 26, 2024

  • 8x

    Median recovery costs were $3 million for organizations whose backups were compromised, eight times the $375,000 for those whose backups survived.

    Source: Sophos, “The impact of compromised backups on ransomware outcomes”, March 26, 2024

    The source doesn't say whether these recovery costs include ransom payments.

  • 93%

    93% of cyber attacks in Veeam's 2023 survey went after backup repositories, and 75% of victims lost at least some of them.

    Source: Veeam, “2023 Ransomware Trends Report”, May 23, 2023

    Survey of 1,200 IT leaders by Vanson Bourne for Veeam, a backup vendor.

  • 28%

    90% of organizations say they're confident they can recover from a cyber incident, but only 28% of ransomware victims whose operations or data were hit fully recovered all affected data.

    Source: Veeam, “Data Trust and Resilience Report 2026”, April 14, 2026

    Survey by Veeam, a backup vendor, of more than 900 IT, security and risk leaders worldwide; the release doesn't say who ran it.

  • 72%

    Organizations recovered 72% of affected data on average after a ransomware attack, and 44% recovered less than three-quarters of it.

    Source: Veeam, “Data Trust and Resilience Report 2026”, April 14, 2026

  • 10%

    Only 10% of organizations hit by ransomware recovered more than 90% of their data, while 57% recovered less than half.

    Source: Veeam, “2025 Ransomware Trends and Proactive Strategies Report”, April 23, 2025

    Survey by Veeam, a backup vendor, of 1,300 organizations, 900 of which had been hit by ransomware in the past year.

  • 57%

    Ransomware victims in Veeam's 2024 survey expected to recover only 57% of their compromised data.

    Source: Veeam, “2024 Ransomware Trends Report”, June 4, 2024

    Survey by Veeam, a backup vendor, of 1,200 executives, security staff and backup administrators.

  • 81%

    81% of organizations in Veeam's 2024 survey paid a ransom, and one in three of those still couldn't recover their data.

    Source: Veeam, “2024 Ransomware Trends Report”, June 4, 2024

  • 42%

    42% of US organizations that experienced data loss were able to restore all of it.

    Source: Backblaze, “2024 State of the Backup”, June 11, 2024

    Survey of 300 US IT decision-makers by The Harris Poll for Backblaze, with a margin of error of 5.7 percentage points.

  • 39%

    39% of US IT decision-makers say their organizations need to restore data from backups at least once a month.

    Source: Backblaze, “2024 State of the Backup”, June 11, 2024

  • 76%

    76% of IT decision-makers in a 2022 global survey reported a severe loss of critical data at their organization.

    Source: Arcserve, September 13, 2022

    Survey of 1,121 IT decision-makers in 11 countries at companies with 100 to 2,500 employees and at least 5 TB of data, run by Dimensional Research for Arcserve, a backup vendor.

How often backups get tested

Most recent surveys on backup testing were run by backup vendors.

  • 15%

    15% of IT professionals in Kaseya's 2025 survey test their backups daily, and another 25% test weekly.

    Source: Kaseya, “State of Backup and Recovery Report 2025”, February 6, 2025

    Survey by Kaseya, a backup vendor, of more than 3,000 IT professionals worldwide; the release doesn't say who ran it.

  • 11%

    11% of organizations test disaster recovery daily and 20% weekly, while 12% test only ad hoc or not at all.

    Source: Kaseya, “State of Backup and Recovery Report 2025”, February 6, 2025

  • 25%

    25% of organizations test disaster recovery once a year or less.

    Source: Unitrends, “The State of Backup and Recovery Report 2025”, January 20, 2025

    Unitrends is a Kaseya brand, so this comes from the same survey as the Kaseya figures, not a separate study.

  • 35%

    60% of IT professionals believed they could recover from downtime in under a day, but only 35% actually could.

    Source: Kaseya, “State of Backup and Recovery Report 2025”, February 6, 2025

  • 24%

    95% of IT decision-makers said their company had a disaster recovery plan in 2022, but only 24% had a mature plan that was documented, tested and kept up to date.

    Source: Arcserve, September 13, 2022

    Survey of 1,121 IT decision-makers in 11 countries, run by Dimensional Research for Arcserve, a backup vendor.

  • 52%

    Only 52% of organizations could recover from a severe data loss within 12 hours.

    Source: Arcserve, September 13, 2022

  • 72%

    In a Q1 2025 survey of 285 German small and mid-sized businesses run by data recovery firm DATA REVERSE, 72% said they rarely or never test whether their backups can be restored. Only 28% test at least once a quarter.

    Source: DATA REVERSE, March 27, 2025. Also: pressetext, “Backup-Studie 2025: Drei Viertel aller deutschen KMUs testen ihre Backups nicht regelmäßig”, March 31, 2025

    DATA REVERSE ran the survey itself, online with follow-up phone interviews, and hasn't published its sampling method or question wording.

  • 25%

    25% of UK businesses and 19% of UK charities have a formal incident response plan.

    Source: UK Department for Science, Innovation and Technology, “Cyber security breaches survey 2025/2026”, April 30, 2026

    Official UK statistics; the survey gives no percentage for how many organizations test their recovery.

What downtime costs

What causes data loss and outages

  • 1.73%

    Backblaze's hard drives failed at an annualized rate of 1.73% in Q2 2026, the highest in a while; the lifetime rate is 1.41%.

    Source: Backblaze, “Backblaze Drive Stats for Q2 2026”, September 29, 2026

    Backblaze monitored 359,101 drives and analyzed 354,415 of them.

  • 1.36%

    Backblaze's drives failed at an annualized rate of 1.36% across 2025, down from 1.55% in 2024.

    Source: Backblaze, “Backblaze Drive Stats for 2025”, February 12, 2026

    Backblaze analyzed 344,196 drives across 30 models.

  • 74%

    74% of Americans who own a computer have accidentally deleted important data.

    Source: Backblaze, “2024 State of the Backup”, June 11, 2024

    Survey of 2,058 US adults by The Harris Poll for Backblaze; it counts people who have ever deleted something, not a share of data loss cases.

  • 40%

    In Uptime Institute's 2025 Data Center Resiliency Survey, 40% of operators said their organization had a significant IT outage caused by human error in the past three years. Of those, 58% listed staff failing to follow procedures among the most common causes (respondents could pick up to three).

    Source: Uptime Institute, “Annual Outage Analysis 2025”, May 2025

    397 operators answered; the 58% is a share of the 40% who had such an outage, not a share of all outages.

  • 80%

    80% of data center operators say their most recent outage could have been prevented with better management and processes.

    Source: Uptime Institute, “Annual Outage Analysis 2025”, May 2025

    Only 97 operators answered this question, which also asked about configuration.

  • #1 cause

    Human error is the number one cause of downtime for Global 2000 companies, according to Splunk and Oxford Economics.

    Source: Splunk and Oxford Economics, “The Hidden Costs of Downtime”, June 11, 2024

    The release doesn't give a percentage.

  • 62%

    A human element was involved in 62% of breaches in Verizon's 2026 data, up from 60% the year before.

    Source: Verizon, “2026 Data Breach Investigations Report”, 2026

    Based on more than 22,000 confirmed breaches in 145 countries; it measures breaches, not data loss.

  • 48%

    Ransomware showed up in 48% of breaches in Verizon's 2026 report, yet 69% of ransomware victims didn't pay, and the median payment fell to $139,875.

    Source: Verizon, “2026 Data Breach Investigations Report”, 2026

  • 26%

    In Sophos's 2026 survey, malicious email (26%) and phishing (24%) were the top root causes of ransomware attacks, while exploited vulnerabilities fell from 32% to 18%.

    Source: Sophos, “The State of Ransomware 2026”, September 2026

    The survey asked about attacks in the previous 12 months, most of which happened in 2025.

Real database and cloud incidents

Supabase backup facts

From Supabase's docs and pricing page, as checked on October 17, 2026.

How often people and businesses back up

Statistics we couldn't verify

Some of the most quoted backup statistics have no source anyone can find. We followed each of these back as far as it goes. Please don't cite them as facts.

“60% of companies that lose their data shut down within 6 months”
The earliest version we found is a January 2009 blog post by IDrive, a backup company, which gives no source. It is a different claim from the cyberattack version below, which is the one the National Cyber Security Alliance disowned.
Source: IDrive, January 22, 2009
“60 percent of small businesses that suffer a cyberattack will go out of business within six months”
In 2017 Nextgov traced it to a September 2011 Business Insider article that cited nothing. The National Cyber Security Alliance, often named as the source, said in a statement dated May 8, 2022: "This statistic was not generated from NCSA research, and we cannot verify its original source."
Source: National Cyber Security Alliance, “National Cyber Security Alliance Statement Regarding Incorrect Small Business Statistic”, May 8, 2022. Also: Nextgov, “How a Fake Cyber Statistic Raced Through Washington”, May 3, 2017
“93% of companies that lost their data center for 10 days or more filed for bankruptcy within one year”
It is usually credited to the US National Archives and Records Administration, but the earliest version we found, in the same 2009 IDrive post, names no report or year, and we found no NARA document with this number. The wording has kept changing as it spread.
Source: IDrive, January 22, 2009
“34% of companies don't test their backups / 77% of companies that test tape backups find failures”
Both first appear in the same 2009 IDrive paragraph, without a source. A product page that credited Storage Magazine is gone and gave no year, and since the numbers are about tape backups they likely date from the mid-2000s or earlier.
Source: IDrive, January 22, 2009
“Up to 58% of backups fail during recovery”
It comes from Veeam's Data Protection Report 2021 press release, which says both "58% of backups fail" and "58% of recoveries fail." Veeam's CTO explained that "Backups are ending with errors or are overrunning the allocated backup window, and secondly, restorations are failing to deliver their required SLAs," which is not the same as data that can't be restored.
Source: Veeam, “Data Protection Report 2021”, March 18, 2021
“Up to 60% of backups are incomplete and 50% of restores fail”
Roundup sites credit it to Avast or Gartner. Gartner's 2009 presentation "The Broken State of Backup" has a slide titled "Backup Success Rates Are Not Acceptable" but no 60% or 50% figure we could find, and Avast's page returned an error when we tried to open it.
Source: Gartner, “The Broken State of Backup”, 2009
“Human error causes X% of data loss”
The number changes with the source and the year. The most quoted one, 40% from a 2010 Kroll Ontrack survey, is what respondents believed was the main cause; only 27% could trace their own data loss to human error.
Source: IT Jungle, July 26, 2010

How we built this list

We only kept numbers we could trace to whoever produced them: the survey report, the incident write-up from the company involved, or the official documentation. For every survey we note who ran it and how many people answered.

Several of these surveys were run or paid for by companies that sell backup products. The note under each number says so. Reports from backup vendors that compete with BackupDrill are named but not linked.

BackupDrill sells backups and restore drills for Supabase projects, so we have an interest in this subject. We kept every number as its source states it, including the ones that don't help our case.

Cite this page

BackupDrill. "Backup and restore failure statistics (2026)." https://backupdrill.com/research/backup-restore-failure-statistics. Updated October 17, 2026.