Moving CRM without losing your history

Most migrations fail in the same place: nobody decided what “the same contact” means before the import ran. Here is the order that avoids the mess.

The BittaFlow team

3 min read

On this page
  1. Why migrations stall
  2. Step 1: export everything
  3. Step 2: decide what a duplicate is
  4. Step 3: map fields, not columns
  5. Step 4: test with twenty records
  6. Step 5: import, then freeze the old system
  7. What to leave behind
  8. Where BittaFlow fits

The software is not the hard part of switching CRM. The data is.

Every failed migration I have seen had the same shape. A file was exported, imported, and then someone opened the new system and found three versions of the same customer and no notes attached to any of them.

Why migrations stall

Three reasons.

Nobody owns it. “We’ll migrate over the weekend” means nobody has booked the weekend.

Nobody defined a duplicate. Two records with the same phone number and different spellings of the name. The importer does exactly what it is told, including the wrong thing.

Nobody tested. The full file goes in first, and the problems are discovered afterwards, at scale.

Step 1: export everything

Not just contacts. Export:

  • Contacts and companies
  • Notes and activity history
  • Deals or jobs, with their stages
  • Attachments, if they matter
  • Tags and lists

Then open each file and look at it. If a column is empty in the export, it will be empty in the new system.

Step 2: decide what a duplicate is

Pick a rule and write it down before you touch the importer.

  • Phone as the key. Best for service businesses where the phone is how people are reached.
  • Email as the key. Better for B2B, weaker for households sharing an address.
  • Phone first, email second. A sensible default.

Then decide what happens on a collision: keep the newest, keep the one with notes, or merge the fields. Each choice has a cost. Choosing nothing has the largest one.

Step 3: map fields, not columns

Your old export has columns the new system does not need. The new system has fields the old export does not have.

Make a two-column list: old field, new field. Anything without a partner gets a decision: drop it, or put it in notes. Do not let the importer guess.

The notes field is where the odd details survive. Use it.

Step 4: test with twenty records

Pick twenty real records, including the messy ones: a contact with two phone numbers, a record with no email, one with a long note, one that is clearly a duplicate.

Import them. Then check by hand:

  • Does the phone dial correctly?
  • Did the notes survive?
  • Did the duplicate rule behave as expected?
  • Are the tags attached to the right records?

Twenty records will expose every problem the full file would, without creating a mess you have to clean twice.

Step 5: import, then freeze the old system

Import the rest. Then make the old system read-only for everyone except one person, for one month.

That month is the safety net. If something is missing, you can still look. After a month, cancel it. Keeping two systems live for a year is how teams end up with two truths.

What to leave behind

Not everything should move.

  • Dead leads from five years ago, unless there is a reason to keep them.
  • Duplicate records you have been meaning to merge.
  • Fields nobody has looked at since the last admin left.

A smaller, cleaner list is easier to trust. The goal is not a complete archive; it is a working system.

Where BittaFlow fits

BittaFlow imports a CSV, maps the columns, and de-duplicates on the rule you choose, with the source tag kept on every record. The test-with-twenty step is built into how the import is meant to be run: small first, then the rest.

If you are planning a move, do the duplicate rule first. Everything else is easier after that decision.

Useful? Send it to someone who runs the phones.