Local, off-site, and actually tested
A backup you have never restored is a rumour.
Nearly every business we assess has a backup. A striking number have one that has been silently failing for months, or one that ransomware would encrypt alongside everything else. The difference only shows up on the worst day of your year.
What it costs
Backup design and implementation quoted after an assessment; off-site storage billed on data volume. Ongoing monitoring and test restores sit in the $650/month retainer.
The three ways backups fail
It stopped running. Months ago. A drive filled, a credential expired, or an update broke the agent, and the failure notice went to somebody who has left. Nobody checked because it had always just worked.
It was reachable from the network. Ransomware hunts backups before encrypting anything, because that is what makes people pay. A drive plugged into the server, or a share the server can write to, gets encrypted with everything else.
It was never restored. The job reported success nightly for two years. The first real restore reveals it was backing up the wrong volume, or that nobody documented the encryption key. Success in a log is not data you can get back.
What gets configured
A local copy for speed, because restoring a deleted folder should take minutes rather than a download. An off-site copy for the fire, flood and theft case. And immutability on the off-site copy, so that even fully compromised credentials cannot delete or encrypt the history.
That is the modern form of the old 3-2-1 rule, and immutability is the part that matters most now. It is also the part missing from most setups we inherit.
Testing is the whole point
Restores get tested and the result written down, including how long it took. That number is what matters when you are deciding whether to keep waiting or start ringing customers.
The test also catches the boring failures: a database backed up while running that restores corrupt, a share whose permissions do not survive, an encryption key nobody wrote down.
Two questions that decide the design
How much data can you afford to lose — a day, an hour, none? And how long can the business be stopped — a week, a day, two hours?
Tighter answers cost more and there is a real trade. A business that can work off paper for two days needs something very different from one where a stopped system means idle staff and missed shipments. That conversation beats being sold the most expensive option by default.
Ransomware, specifically
If it happens, backups decide whether it is an expensive week or the end of the business. The ones who recover well have an off-site immutable copy, know their restore time because they measured it, and have somewhere clean to restore onto.
The ones who pay are the ones whose backup was reachable from the machine that got encrypted. Nothing else about their security posture matters as much as that single detail.
FAQ
Questions we get asked
How often should backups run?
For most businesses, image-level backups through the working day and a nightly off-site copy. If you can genuinely tolerate losing a day's work, nightly alone is cheaper and that is worth saying.
Is a backup drive plugged into the server enough?
No, and it is actively risky. Ransomware encrypts attached and reachable drives deliberately. You need a copy the compromised machine cannot write to.
Does OneDrive or Google Drive count as a backup?
It is sync, not backup. A file that gets encrypted, corrupted or deleted syncs faithfully in that state. Version history helps for one file and falls apart across thousands.
How do we know the backup works?
Because somebody restored from it recently and wrote down how long it took. Any provider who cannot tell you the date of your last successful test restore is guessing.
Can you recover data after a ransomware attack?
If there is a clean off-site copy, usually. Without one the options narrow to whatever decryptor happens to exist. This is a problem solved beforehand, not afterwards.
Ready when you are.
Ring and describe it in plain English. You will get a number before anyone drives anywhere.