How to Migrate Your Data When Switching Password Managers
The export file is the thing nobody thinks about until it is the only thing that matters. A CSV or JSON sitting in a downloads folder, tens of thousands of characters long, holding every login a person owns. It is the most dangerous file on the machine and the most ignored. Password managers make switching sound like a two-click courtesy. The reality is a small, tense migration project with a real chance of losing an account or leaving a credential exposed on a hard drive.
Most people switch managers for reasons that arrive suddenly: a renewal price jumps, a breach notice lands, a company gets acquired. The decision takes an afternoon. The migration takes longer, and the mistakes in it tend to surface weeks later, when a login fails at the worst possible moment.
The work divides into three phases that should never be collapsed: getting the data out cleanly, getting it in cleanly, and destroying the intermediate copy. Each phase has its own failure mode.
What the export file actually contains
An export is not a tidy list of usernames and passwords. Depending on the manager, it holds URLs, login names, passwords, notes, custom fields, folder assignments, two-factor backup codes, and timestamps. Some managers export secure notes as separate records. Some flatten everything into a single file with a column for type. Others produce an encrypted archive that needs the old app to decrypt before it becomes readable.
That structure matters because the receiving manager has to interpret it. A field labeled "notes" in one export may contain a recovery phrase, a PIN, or the answer to a security question. Import tools guess. When they guess wrong, data lands in the wrong field, and a credential that looks present is actually unusable.
Before touching anything, open the export in a plain text editor and read the header row. Count the columns. Note which ones carry the actual secrets and which carry metadata. That five-minute read prevents most of the confusion that follows.
The plaintext window is the real risk
Here is where the process earns its danger. Almost every password manager exports to an unencrypted format by default, because the whole point is portability. For the minutes or hours between export and secure deletion, every password sits in readable form on a disk.
That file should live in exactly one place, briefly, and then stop existing. Not in a synced cloud folder. Not on a desktop that backs up automatically. Not attached to an email draft as a personal reminder. A local folder, a short lifespan, and a deliberate deletion, ideally with a secure-delete tool that overwrites the sectors rather than just unlinking the file name.
The export should also be treated as compromised the moment it is created. Anyone who reads it has every credential. A reasonable practice is to rotate the passwords that matter most after the migration settles: banking, primary email, and anything tied to account recovery. That is tedious. It is also the only way to be sure the plaintext window closed cleanly.
Import into the new manager before deleting anything
The instinct to clean up the old vault first is backwards. The old manager stays intact and untouched until the new one has been verified. Import the file, let the tool map fields, and then inspect the result by hand.
Inspection means opening a sample of entries across categories. A dozen logins, a couple of secure notes, a payment card if the manager stores them, and any entry with a custom field. Check that the username arrived in the username field and not in the notes. Check that the URL is present so autofill will actually trigger. Check that the item is in the right folder or collection.
Duplicates are the other common artifact. Import tools often create a second copy of an entry that already existed, sometimes with a slightly different name. A quick sort by name reveals clusters of near-identical records. Merging them by hand is dull work, but leaving them creates confusion at exactly the wrong moment, when a login prompt offers two similar entries and the wrong one autofills.
Why one export is usually not enough
Password managers fragment their data in ways users forget. There may be a personal vault and a shared vault for a household or a team. There may be a separate store for passkeys, which many managers now handle alongside passwords but export inconsistently, or not at all. There may be attachments, such as scanned recovery documents or images of backup codes.
A single export often captures only the main vault. The shared items and the passkeys sit outside it, and they are exactly the credentials people rely on daily without thinking. Passkeys in particular have weak portability across managers right now, and some ecosystems tie them to the platform rather than the password manager itself.
The practical move is to inventory what the old manager holds before exporting: how many vaults, whether shared collections exist, whether passkeys are stored, whether attachments are in use. Then confirm against the export that each category made it into the file. What is missing has to be recreated by hand, and it is better to know that on a quiet evening than during a login attempt.
Two-factor codes and recovery paths need their own check
Two-factor authentication is a separate layer that people mentally fold into the password manager. It is not the same thing. If the old manager stored TOTP codes, those seeds may or may not be in the export, depending on the tool and the export format. If they are not, every account with code-based two-factor will demand a fresh setup.
That means keeping the old authenticator, or the old manager's app, alive until every account has been re-enrolled. Deleting the old app early is how people lock themselves out of their own email. The old device becomes a bridge, and it should stay standing until the far bank is reached.
Recovery codes deserve the same care. Printed backup codes, security questions, and recovery email addresses all need to point somewhere valid. A migration is a good moment to confirm that the recovery email on the most important accounts is one the person still controls. Old recovery addresses are a quiet liability that outlives any password manager.
The old vault should be retired, not abandoned
Once the new manager is verified, the old one should be dealt with deliberately. Abandoning it means an account stays active, possibly still paid for, still holding a full copy of every credential, and still reachable if the old master password leaks. The clean sequence is to delete the vault contents, then close the account, then confirm by trying to log in and finding it gone.
Some services make this harder than it should be, burying account deletion behind a support request or a retention period. A calendar reminder for the follow-up is worth setting. An old vault that was supposed to be deleted and quietly was not is a worse outcome than never switching at all.
The export file itself gets the same treatment. Secure deletion, then a check that no synced copy survived in a backup service or a cloud drive's version history. Those services keep deleted files for weeks. The plaintext window is only truly closed when the backups have rolled past it.
A dry run on one device before the full switch
The last habit worth adopting is to install the new manager on a single device first and live with it for a day or two before migrating every other device. Autofill behavior, browser integration, and mobile keyboard support all vary in ways no spec sheet captures. A manager that feels fine on a laptop can be clumsy on a phone, and the phone is where most logins happen.
That trial period also surfaces the entries that imported badly. A login that fails on one device is a signal to check the field mapping before trusting the rest of the vault. Finding three broken entries during a quiet trial is a minor annoyance. Finding them during a Monday morning meeting, locked out of a work account, is a different kind of day.
Migration is unglamorous work, and the tools do most of it. The residual human tasks are the ones that decide whether the switch holds: reading the export, watching the plaintext window, verifying a sample by hand, keeping the old bridge standing until the new bank is reached, and then tearing the bridge down completely.


