Backup vs Archive And Why One Cannot Do Both
The organisation believed it had seven years of mail. It had fourteen days, repeatedly, for seven years.
A backup is for recovery and an archive is for discovery. Those four words carry most of it. A backup answers one question. Something broke, put it back the way it was. An archive answers a different one. What did we have in 2019. Can one message be found inside it. That is a searchability question rather than a restore one. The two are built for those questions, which is why using either as the other quietly fails.
This matters most in mail, where people say backup and mean whichever copy they happen to have. A mailbox export, a synced folder, a nightly job. Those three behave nothing alike on the day somebody needs something back.
Backup And Archive, Two Different Questions
Start from what each one is for and the rest follows.
A Backup Exists For
- A disk that failed this morning
- A folder somebody deleted yesterday
- Ransomware that arrived overnight
- Getting back to how things were
An Archive Exists For
- A request about a contract from 2019
- An employee who left four years ago
- A regulator asking for seven years
- Finding one thing among millions
The clearest test is how you get things out. From a backup you restore the whole thing to a point in time and rarely go hunting for one item. From an archive you search for one thing and take only that. Restoring the lot would be useless. Any copy you cannot search is a backup whatever you call it.
Why a Backup Cannot Hold History
This failure costs people the most. Nobody made a mistake to cause it either, because it is built into how backups work.
Every backup scheme reuses its storage. Yesterday overwrites the day before. This week overwrites last week. The cycle is the point. It keeps cost and storage bounded.
What a rotation does
Working exactly as designedday 1 written
day 2 written
...
day 15 overwrites day 1
day 16 overwrites day 2
anything older is goneNothing failed here. The scheme is doing the job it was configured to do.
What that means for history
A deletion on a timer2019 mail
in a 2019 backup
overwritten in 2019
asked for in 2026
nothing to search
nothing to restoreTreating backups as your archive builds a rolling deletion into your own records that nobody signed off on.
Long retention on a backup does not turn it into an archive. Keeping ninety days rather than fourteen buys time and changes nothing structural. You still cannot search it, still restore whole systems to get at one message and still lose everything past the window.
Where Synchronisation Sits Against Backup And Archive
Neither. It gets called both more often than either of them.
A synced mailbox or drive copies whatever the source holds, which includes every deletion. Remove something by accident and the removal reaches your phone before you have finished regretting it. That is the feature working.
| What goes wrong | Sync | Backup | Archive |
|---|---|---|---|
| Laptop stolen | Covered | Covered | Covered |
| You delete a folder by mistake | No, it syncs the deletion | Covered | Covered |
| Ransomware encrypts everything | No, it syncs that too | Covered if offline | Covered |
| Somebody asks for 2019 | Only if never deleted | No, rotated away | Covered |
| Account closed by the provider | No, every copy stops | Covered | Covered |
Read the sync column downwards and the shape shows. It handles hardware trouble and nothing else. Every other row involves the data itself changing. Copying changes faithfully is the job.
An IMAP mailbox is synchronisation, not backup. Every device holds a copy and the server decides what all of them contain. That is worth knowing before treating a mail account as the place your history lives. It is a common reason mail turns up missing with nowhere to get it from.
Read next IMAP or POP And Which One You Already Use Which of your mail copies is a mirror rather than a safety net.What Backup And Archive Mean for Mail Specifically
Mail makes the distinction sharper than most data does, because the requests that arrive are almost always about one message from years ago.
| What you have | What it actually is |
|---|---|
| A mailbox export made once | An archive, if you keep it and can search it |
| A nightly server job | A backup, with a rotation window |
| Mail on your phone and laptop | Synchronisation, which is neither of the others |
| Deleted items retention | A short grace period, measured in days |
| A PST somebody made in 2019 | An archive, whether or not anybody planned it |
That last row is how most real archives come about. Somebody exported a PST before leaving a job and it sat on a drive. Nobody designed it and it is the only reason 2019 still exists.
An accidental archive still has to be readable. A file nobody can open is not an archive, whatever it holds. That is the argument for keeping such things in a format anything can read, rather than one tied to a program that may not be installed anywhere by the time somebody asks.
Building a Backup And an Archive Without Building Twice
- Decide the two windows separately. One is how far back you might need to undo. The other is how far back somebody might ask you to look. The first is usually weeks. The second is usually years. They are different numbers for different reasons.
- Let the backup do the short window. Frequent, automatic, overwritten on a rotation. Keep at least one copy where nothing running can reach it.
- Make the archive on purpose. Export at a point in time, put it somewhere nothing writes to by accident, in a format that will still open.
- Check the archive by opening it. Not by confirming the file exists. Open it, search it, find something specific. An untested archive is a hope.
- Write down what each one covers. The failure is nearly always somebody assuming a copy answers a question it was never built for.
The one question worth asking about any copy you hold. If somebody asks for a single message from four years ago, can you find it without restoring anything else. Yes means archive. No means backup. Not sure means neither, yet.
Where an old export needs checking rather than trusting, our file viewers open one in a browser tab so you can search it before relying on it. Where the format is the obstacle, the converters move the contents into something current. The backup tools handle pulling mail down from an account while you still have one.
Definitions and retention behaviour checked against published reference material in August 2026. Retention rules vary by jurisdiction and industry, so treat the legal points as orientation rather than advice.
Questions People Ask
7 questions, answered in full below.Can my backups serve as an archive?
Only for as long as the oldest one survives, which is usually weeks. Backup schemes reuse their storage on a rotation, so anything older than the cycle has already been written over. An archive that deletes itself on a timer is not doing the job.
Is a synced folder a backup?
No. Synchronisation copies deletions as faithfully as it copies files, so a mistake reaches every device in seconds. It protects you from losing a laptop. It protects you from nothing you do to your own data.
How long should a backup be kept?
Long enough to notice the problem it exists for. Most damage is found within days, which is why a fortnight is a common answer. Anything you want in five years belongs somewhere else, under a decision made on purpose rather than left to a rotation.
Does Microsoft 365 back up my mail?
It keeps your mailbox available and it replicates it, which is not the same thing. Deleted items have a recovery window measured in days rather than years. Anything you need beyond that window is your responsibility. A great many people discover this at the wrong moment.
What is the difference in practice, not theory?
How you get things out. From a backup you restore the whole thing to a point in time. Looking for one item is rare. From an archive you search for one thing and take only that, since restoring everything would be pointless.
Do I need both?
If losing last week would hurt and losing 2019 would also hurt, then yes, because no single copy answers both. Most people who think they have both have two backups, one of which they call an archive.
Where does an archive actually live?
Somewhere nothing writes to it by accident. A drive kept offline works. So does storage that refuses changes once written. So does a service built for retention. The requirement is not the medium. It is that no ordinary process can quietly alter what is there.
Sources
Where the figures and behaviour described above were checked.
- Backup Wikipedia
- Recovery point objective Wikipedia
- Legal hold Wikipedia