Cloud Data Loss Survival Checker: If Your Provider Lost It All
Suppose the provider that holds each of your data sets lost all of it, permanently. What would you still have, and how fast could you get it back? To our knowledge no major cloud provider has ever lost all of its customers' data, so treat this as a resilience exercise, not a forecast. Outages like the Microsoft 365 outage that began on 12 November 2026 are a good prompt to check where your only copies live.
Shared responsibility, in plain words: SaaS and cloud providers generally keep the service running, while protecting against the loss of your own data (deletion, ransomware, a compromised admin, a contract ending) is often the customer's job. What your provider backs up, and for how long, varies by product and contract: check your provider's terms rather than assuming.
1. Your data, by provider
For each data set, name the provider that holds the live data and answer as things are today. Untick anything you do not have. Times are in hours (a day is 24). A data set counts once you answer the first question.
Answer “Independent copy outside that provider?” for at least one data set to see your result. Nothing you type is sent to us.
How the score works
A transparent PlayCISO heuristic for this exercise, not a standard or a benchmark. Every point is listed here.
- Each data set scores 0 to 100. A copy outside the provider that holds the live data earns 40 points. Without one the data set scores 0, because in this scenario nothing held by that provider survives (a copy in another region or another service of the same provider counts as "same provider").
- If there is an outside copy: +15 if it sits behind separate admin accounts and MFA (not the same identity or single sign-on as the live data), +15 if at least one copy is immutable or offline, +20 for a test restore in the last 90 days (+12 within a year, +5 longer ago, 0 never), +5 if copies are taken at least as often as the data loss you can tolerate (RPO) and +5 if the estimated restore time fits the outage you can tolerate (RTO).
- "Not sure" scores like "no": an unconfirmed control is not one you can count on in a crisis. A blank time scores 0 for that check.
- The overall score is the average of the data sets weighted by criticality (critical 4, high 3, medium 2, low 1). Rows you untick, or leave unanswered, are not counted.
- Bands: 80 and above "Likely to survive", 50 to 79 "Partial survival", below 50 "Severe loss". These are PlayCISO thresholds for this exercise, not an industry standard.
Your answers, the score, the lost and single-point-of-failure lists, the restore order and the fixes, ready for a risk register or a board note.
TL;DRWhat this checker does
4 points
- List each data set (mailboxes, files, chat, identity, code, databases, VM images, logs, password vault, contracts and HR records, or your own) and the provider that holds it.
- For each, say whether an independent copy exists outside that provider, whether it is immutable or offline, behind separate admin accounts, when a restore was last tested, and your RPO, RTO and restore time.
- You get a survival score with every point explained, the data you would lose forever, backups that share a provider or an identity with the live data, a restore priority order and a fix list.
- Free to use, signed out too. It runs in your browser and nothing is sent to us. Markdown and CSV exports are available with a paid plan.
Frequently asked questions
Does my cloud provider back up my data?
It depends on the service and your contract, so check your provider's terms rather than assuming. Many SaaS and cloud agreements follow a shared responsibility idea: the provider keeps the service running and its infrastructure resilient, while protecting your data against your own deletions, ransomware, a compromised admin or a long-term loss is often the customer's job. Features like a recycle bin, version history or replication are not the same as an independent backup you control.
What is the shared responsibility model for SaaS backup?
It is the split of who protects what. In plain terms, the provider is generally responsible for running the service, and the customer is generally responsible for their data, identities, access and configuration. Where exactly the line sits differs by provider and product, so read your provider's own documentation and your agreement. This checker assumes the worst case on purpose: the provider has nothing left, so only copies you hold elsewhere count.
Has a cloud provider ever lost all of its customers' data?
To our knowledge, no major cloud provider has ever lost all of its customers' data. Outages are far more common than permanent loss. This checker is a resilience exercise: picturing the extreme case is a quick way to find data that exists in one place only, or behind one login.
Does a backup in another region of the same cloud count?
Not in this exercise. A second region, a recycle bin, versioning or the provider's own backup add-on all help against many failures, but they sit with the same provider and often behind the same admin accounts. The checker counts them as "only with the same provider" and lists them as a single point of failure.
Why does the identity used for backups matter?
If the backup console signs in through the same identity provider, single sign-on or MFA as the live data, losing or compromising that identity can lock you out of the backup too. Keep backup admin accounts separate, with break-glass credentials stored offline, and restore identity and the password vault first, because every other restore needs someone to log in.
How often should I test a restore?
There is no single rule; set it per system from its criticality, RPO and RTO. The checker gives the most points for a successful test restore in the last 90 days and none for "never", because an untested backup is a hope, not a recovery plan. Time the test so your restore estimate is a measurement, not a guess.
More security tools
The tools people use most on PlayCISO.
Catch injected instructions hiding in content your AI agent should only read.
Score your readiness against the first international standard for AI security threats.
Critical CVEs across 21 major vendors, month by month — and what the 2026 surge means.
Build an ordered ruleset, replay labelled packets and catch shadowed and any-to-any rules.
Self-assess all six Functions, from Govern to Recover, and see your weakest one.
Search 172 threat actors across Microsoft, CrowdStrike, Mandiant and community names.