If you have backups, can you actually restore the app?

I have a fairly strong opinion about backup dashboards: a green status is comforting, but it tells you surprisingly little about whether the business can recover.

You might have every virtual machine backed up on schedule and still discover that nobody has the database credentials, the restored service cannot talk to its dependencies, or the backup storage is painfully slow when you need to recover several systems at once.

The question is what comes back first

If an important application went down tonight, what would you restore? In what order? Where would it run? How long could users reasonably be without it? Somebody should be able to answer those questions without reading twenty pages of old disaster recovery documentation.

The gap between confidence and reality is not hypothetical. Veeam’s 2026 resilience survey found 90% of security leaders confident they could meet recovery time targets, but only 69% said those targets fully matched business continuity goals.

I would pick one real application and do a recovery exercise outside the production environment. Record the missing steps, fix the credentials and dependencies, and repeat it. That is less impressive in a presentation than a new backup platform, but a lot more useful when things go wrong.

Try restoring something that matters

I'd pick an application people would genuinely miss and walk through a recovery exercise. Not a screenshot of a successful backup job. A real restore into an isolated environment, with the database, application files, secrets and dependencies needed to make it useful.

Pay attention to the little questions that come up. Where are the credentials stored? Does the restored application know how to reach its database? Are the backups protected from the same account that could compromise production? Who can make the decision to switch users to the recovered environment?

Recovery time targets are useful only when they survive contact with reality. If the business expects an hour but a test restore takes six, you've learned something important while everyone is calm. That is much cheaper than learning it during an outage.