
What Auditors Actually Ask About Your Backups
Auditors ask about backups in a predictable order: show me the policy, show me the schedule, show me that failures get noticed, and show me a restore you actually performed.
Auditors ask about backups in a predictable order: show me the policy, show me the schedule, show me that failures get noticed, and show me a restore you actually performed. We answered every one of those in a real IT general controls (ITGC) audit this year. Below is the question sequence as we experienced it, and the kind of evidence that satisfied each.
Why this post exists
Almost everything written about audit preparation is written by auditors marketing services or consultants selling engagements. This is the other side of the table: this year our company went through an ITGC audit conducted by a Big-4 firm, and the backup and recovery area was examined in depth. What follows is the shape of that examination in our own words (paraphrased areas, not the auditor's document), because knowing the order the questions come in is most of the preparation.
The predictable order
Backup questions in an ITGC audit run policy-first, evidence-last, each answer setting up the next request:
- "Show me the backup policy."
The opening move, because everything after is measured against it. What matters: the policy exists, has an owner and a review date, and, critically, describes what you actually do. A policy that promises daily backups you run weekly converts a process question into an integrity question. If you don't have one, start with our sample policy.
- "Show me what is backed up, and how often."
The coverage question. The auditor has your system list from earlier audit phases and will pick systems from it: where is this one backed up? The evidence that works is a backup register (system, method, destination, frequency, retention) that agrees with both the policy and the live configuration. The classic failure found here is the system added last year that never joined the backups.
- "How would you know if a backup failed?"
The monitoring question, and the one that separates paper compliance from a working control. "The software sends an email" is a weak answer if the next question (show me a failure and what happened next) has no answer. Counter-intuitively, a clean history is less convincing than an alert trail showing a failure noticed, ticketed and fixed: detection you can demonstrate is the control.
- "Show me a restore you actually performed."
The centrepiece. Not "could you restore?" but "show me one you did: the date, the system, who did it, how long it took, and how you confirmed the result was good." A dated restore-test record with a verified outcome satisfied this; a green job log would not have. If you do nothing else before an audit, perform one real restore test and write it down properly.
- "How long are backups kept, and does anything enforce that?"
Retention: the policy number is compared against the tool's live configuration. Drift between the two (policy says 12 months, software says 30 days) is one of the most common findings in this area, because retention gets adjusted during disk-space emergencies and never reset.
- "Who can access (or delete) the backups?"
Access control, with a modern sharp edge: ransomware. Expect interest in whether backup credentials are separate from everyday admin accounts, and whether any copy is immutable, protected from deletion even by an administrator. "Everyone in IT can delete backups with the same domain admin account" is a finding.
- "Where is the offsite copy?"
The disaster question, usually last: what survives if the building doesn't, and how recovery would proceed, which is your DR plan, with the backup evidence as its foundation.
What we learned the hard way
Three honest lessons from the audited side:
The asking is iterative. Every answer produces a follow-up request for the evidence behind it. Answering "yes we test restores" means the next email asks for the records. Assume every claim will need an artifact, and you will draft your answers differently.
Scattered evidence costs more than missing evidence. Facts that lived in mailboxes, chat threads and someone's memory took far longer to assemble than facts that lived in a system designed to be shown. The audit is largely a retrieval exercise; you win it before it starts, by keeping evidence where it can be produced in minutes.
Honesty about gaps works. A gap you present yourself, with an owner and a date, reads as a functioning control environment. A gap the auditor discovers reads as the opposite, even when it's the same gap.
Preparing in a week vs. a quarter
Triage by time available
| Time available | Do this, in order |
|---|---|
| One week | Adopt a policy that matches reality (§1); build the backup register (§2); perform and document one real restore test (§4). Those three answer more than half the examination. |
| One quarter | All of the above, plus: working failure alerts with a reviewed trail (§3); align retention config to policy (§5); separate backup credentials and add an immutable copy (§6); confirm the offsite copy and write the DR plan (§7). |
The through-line: every question above is answered by evidence that a system can generate continuously: verification records, alert trails, restore tests with dates and durations. That is precisely the evidence Backup Governor accumulates as a by-product of doing its job, which is why we built it the way we did: we had just sat this exam. The wider evidence picture is in How to Prove Your Backups Actually Work.
CTA
Walk into the audit with the evidence already collected
Verification history, alert trails and dated restore tests, accumulated automatically and ready to show.
Request a demo