The short answer
- Artwork, customer records and order history are irreplaceable; everything else can be rebuilt
- Hosted software is backed up by them, for them — that is not the same as a copy you hold
- A backup you have never restored from is a hypothesis
- Keep tomorrow's jobs printable on paper, so an outage costs hours not days
- Export on a schedule, and keep at least one copy somewhere else
Decorating shops think about equipment failure and rarely about data failure, which is backwards in terms of consequence. A press that stops costs you a week of production. A customer list and ten years of artwork that vanish cost you the business's accumulated advantage, and no insurance replaces them.
What is actually irreplaceable?
| Data | If lost |
|---|---|
| Customer artwork and separations | Cannot be recreated — must be redone at cost |
| Customer list and contacts | Partially recoverable, painfully |
| Order history | Gone; reorders become new jobs |
| Approved proofs | Gone; disputes become unwinnable |
| Price list | Rebuildable in a day |
| Product catalogue | Re-syncs from suppliers |
The top four are the priority and they are all customer-linked. Everything below can be reconstructed from other sources with effort. That ordering should determine what you protect most carefully rather than treating all data as equivalent.
Approved proofs deserve particular attention because their value is entirely evidential. A shop that cannot produce the proof a customer approved has lost the only thing that settles a disagreement about what was agreed.
Hosted does not mean safe
If your system is hosted, the provider almost certainly backs it up. That protects them against their own hardware failing. It does not protect you against three things.
Your own mistakes. If somebody deletes a customer and the deletion replicates to the backup, the provider's redundancy has faithfully preserved the deletion.
Losing access. A billing lapse, a contract dispute, a provider ceasing to trade. Your data may exist and be unreachable, which is functionally the same as gone.
Wanting to leave. The ability to export is a product decision the vendor made, and it varies enormously. That is why a real sample export you can open is worth asking for before signing rather than after.
Their backup protects their service. Yours protects your business. They are different things.
Testing a backup
The only meaningful test is restoring from it. Confirming that a backup job ran successfully tells you a file was written, not that the file is usable.
The restore test, once a quarter
- Pick a specific customer and a specific job from a year ago
- Restore only that, to somewhere separate
- Open the artwork — does it open, in the right application?
- Are the separations there, and the approved proof?
- Is the order data readable, with sizes and specification intact?
- Time it. How long would a full restore take?
The artwork check catches the commonest silent failure. Exports frequently include images but not the layered working files, or produce files stripped of their names — technically a backup, practically useless for producing a reorder.
Timing matters too. A restore taking three days is a different business risk from one taking two hours, and knowing which you have changes what contingency you need.
Printing through an outage
Systems go down — an outage at the provider, a connection failure, a power cut. The question is whether that stops production or merely inconveniences it.
What keeps the presses running
- Tomorrow's jobs printed on paper at the end of each day
- Print-ready artwork for scheduled jobs held locally, not only in the cloud
- A phone list of suppliers and key customers, off-system
- A way to take payment that does not depend on your main system
- An agreed manual process for recording what was produced
The first line is the one that turns a crisis into an inconvenience. A shop that starts each day knowing what it is printing can print it regardless of what the network is doing, and reconcile afterwards. A shop where the schedule exists only on a screen stops entirely.
The reconciliation matters — record on paper what was actually produced, so the system can be updated once it returns. Otherwise the outage produces a second problem: jobs that were made but never marked, which surfaces later as a customer asking where something is that has already shipped.
The paper fallback
One page per job, printed the day before. It feels antiquated and it is the cheapest insurance in the shop.
| The job sheet carries |
|---|
| Customer, order number, contact |
| Garment: brand, style, colour, size breakdown |
| Decoration method, colours, print size and placement |
| The approved proof, printed |
| Due date and delivery method |
| Any special instruction — names, packing, exchanges |
Most shops already produce something like this. The discipline worth adding is that it is generated in advance rather than on demand, so it exists when the system that generates it does not.
Ransomware and access
Small businesses are targeted precisely because they are unlikely to have tested backups. Three measures cover most of the realistic risk.
Keep one copy offline or immutable. A backup on a drive permanently connected to the machine it backs up will be encrypted along with everything else. One copy that cannot be written to by a compromised system is what makes recovery possible.
Individual logins. Shared credentials mean no audit trail and no way to revoke one person's access. That is a permissions question before it is a security one, and the reasoning is the same either way.
Remove access promptly when people leave. Dormant accounts belonging to people who no longer work there are the most common route in, and closing them is part of the same routine as collecting keys.
None of this is elaborate. Export on a schedule, keep a copy somewhere else, restore from it once a quarter, and print tomorrow's jobs tonight. That is an afternoon to set up and it is the difference between a bad week and a business that cannot answer what it owes whom. The rest of the software archive assumes you have restored from your backup at least once.


