Why backups fail exactly when needed
Backups fail quietly and in advance: the backup job that has been erroring for months, the copy that saved the pages but not the customer data behind them, the backup stored on the same machine that just died, the restore that technically works and produces your site as it was in 2023. None of these announce themselves. They are discovered during the emergency, which is the one moment you cannot fix them.
The standard that matters: restored, not taken
A backup only counts once the restore has been practised. The standard worth holding any provider to: backups restored somewhere safe on a schedule, with the result checked by a machine rather than assumed. That turns "we have backups" from a claim into a fact that gets re-proven every morning. It is why the checklist lists restore-testing, not backups, as the daily item.
What good looks like for an Irish small-business site
- Daily automatic backups of the pages and the data behind them, together.
- Stored elsewhere: a backup living on the machine it protects shares that machine's fate.
- Restore tests on schedule, with the result checked automatically, not assumed.
- Copies going back weeks: a hacked site needs a backup from before the break-in, which might be further back than yesterday.
- A written restore time: knowing recovery takes 40 minutes changes how scary everything else is.