How to Migrate Your Existing Data Into a New App Without Losing It

Why migration deserves its own plan

Existing data is messy in ways you stop noticing: names spelled three ways, dates in mixed formats, notes in the wrong column, records nobody has touched in years. A new app has stricter rules, so those problems surface immediately. Treat migration as a mini-project with an owner, a timeline and a test, not a final task squeezed in the night before launch.

Step 1: Decide what to bring

Not everything needs to move. List each data source and choose one of three fates:

Moving less data makes the project faster and the new system cleaner. If in doubt, archive.

Step 2: Audit your data

Open each file and note what you find. Answer these questions:

• How many records are there, and how many are active?
• Is there a unique identifier for each record, such as a client number or email?
• Which columns are always filled, and which are patchy?
• Are there duplicates or near-duplicates?
• Do dates, phone numbers and amounts follow one format?
• Do fields contain more than one piece of information, such as “John Smith, john@email.com”?
• Does any data include sensitive or personal information that needs special care?

Step 3: Clean before you move

Cleaning in the old file is far easier than fixing in the new system.

1. Remove or merge duplicates. Decide which record wins.
2. Standardise formats: one date format, one country spelling, one phone style.
3. Turn free text into choices where possible. “Status” should be a short list, not whatever people typed.
4. Split combined fields into separate columns.
5. Fill critical gaps, such as missing emails for people who will receive notifications.
6. Add a unique ID column, if there is not one, and keep it for traceability.

Step 4: Map old fields to new ones

A field map is a simple table linking every column in your old data to a field in the new app. It is the single most useful migration document.

This is also the moment to confirm the data model: what things exist, how they relate and who can see them. Read about getting your data model right from the start. Relationships matter: if a client has many projects, the import must link each project to the right client, usually through a shared ID or email.

Step 5: Test import on a small sample

1. Choose 20 to 50 records that include typical, awkward and edge-case examples.
2. Import into a test copy of the app, never into live data.
3. Check each record against the original: fields, links, dates, files.
4. Fix the map or the data, then repeat until the sample is clean.
5. Import the full set into the test copy and run counts and spot checks.

Bubble.io supports CSV uploads and bulk changes through workflows, and large imports are best run as controlled batches to protect performance and cost. See performance optimisation.

Step 6: Run in parallel, verify and switch

1. Freeze a cut-off date. Decide when the old system stops accepting changes.
2. Do the final import of the latest data.
3. Verify with a checklist (below) and have a second person confirm.
4. Run both systems for a short time, comparing results on real tasks.
5. Switch over, make the old file read-only and keep a backup.
6. Tell users what has changed and where to report anything missing.

The verification checklist

Privacy and security during migration

• Share data files through secure channels, not casual email attachments.
• Share only what the studio needs, and use anonymised samples for development.
• Delete temporary copies when you are done.
• Confirm that privacy rules are in place before real data goes in. See security best practices.
• If you hold personal data, check your obligations under the data-protection laws that apply to you.

Common mistakes

• Starting the data work after the app is built.
• Importing without cleaning, then living with the mess.
• Not keeping a unique ID, which makes later fixes hard.
• Importing straight into live data.
• Forgetting attachments and files.
• Leaving no time for the final verification.

Where the old system is a spreadsheet, see also signs your spreadsheet has outgrown itself for the wider move.

Frequently asked questions