FillTrust

How to answer

Are backups tested by performing restores?

Everybody takes backups. This question is about the last time one was proved to work.

A model answerImplied, not stated

Yes. Backups are taken [daily], retained for [30] days, and stored in a separate region from production. A restore is performed and verified at least [quarterly], and the result is recorded.

Square brackets are yours to fill in. FillTrust grades an answer like this implied, not stated when your documents support it.

Answered from
Your business continuity plan, and the availability section of your SOC 2 report.
Evidence to attach
Occasionally a record of the most recent restore test, with its date and outcome.
Where it is asked
CAIQ v4.0 · BCR-08SIG LiteISO 27001 Annex A · A.8.13

There is a reason this question is phrased around restores rather than backups. An untested backup is a belief, and the failure mode it protects against, discovering at the worst possible moment that the file is empty or the schema has drifted, is common enough that reviewers ask about it specifically.

The honest answer for most small teams is uncomfortable. Backups run automatically, nobody has restored one deliberately, and the closest thing to a test was a production incident that happened to end well. If that is the position, the useful answer says backups run daily with a stated retention, and that restore testing is being formalised. That is weaker than a yes and much stronger than a yes you cannot support, because this is one of the questions a reviewer may come back to with "when was the last one".

Making it a real yes is a small job. A quarterly restore into a scratch environment, with the date and the outcome written down, takes an afternoon to set up and converts this question, permanently, into a confirmed answer. It is one of the highest-value hours on this entire list.

Answer the three things underneath it. Frequency, retention and location. Daily, thirty days, and a different region from production is a normal shape. If your backups sit in the same region as the thing they protect, say so rather than implying geographic separation you do not have: it is a legitimate design choice for many products and an outright misstatement if you claim otherwise.

Point at your recovery objectives if you have them. A restore test means more alongside an RTO and RPO, because it shows the test was measured against something. If your business continuity plan states four hours and one hour, and your last restore took ninety minutes, that is a very strong two-sentence answer.

FillTrust usually grades this as implied, and it is the right grade: your continuity plan almost certainly mentions backups and almost certainly does not mention testing them. That gap is the answer worth writing by hand once, after which it is reused everywhere.

How this one goes wrong

Specific to this question, not general advice.

  • Answering for backups when the question is about restores. Every team backs up. Far fewer have ever restored one on purpose.
  • Counting a restore into staging you did once, eighteen months ago. The question implies a cadence.
  • Not knowing the retention period or whether backups leave the production region, both of which are immediate follow-ups.

There are another two hundred of these in the file.

FillTrust drafts every one from your own documents and shows the passage behind each answer, including the ones it refuses to answer.

Or write this answer down once and publish it on a Trust Center of your own, which costs nothing.