Moving to a New CRM Without Losing Data

Export everything, including history

The temptation is to export contacts and deals and leave the rest. History is what makes a CRM useful on day one in the new system: the note that says why a deal stalled, the email thread with the decision-maker. Export activities and notes with their record IDs so they can be re-attached, and export files or at least a list of them. Export the old system's own configuration too, stages, custom fields, automations, because it is the spec for what the new one must reproduce.

Clean before you move

A migration is the one time you get to fix years of accumulated mess without anyone objecting. Do it before import, in a spreadsheet or a dedicated cleaning tool, where mistakes are cheap to undo:

• Deduplicate contacts by email and companies by organisation number or domain; decide the merge rule before running it.
• Standardise formats: phone numbers, country, postcode, company names with one spelling.
• Fill or flag required fields the new system will insist on, such as deal value or stage.
• Archive rather than migrate: contacts with no activity in years, and personal data you no longer have a reason to keep under GDPR.
• Assign owners for every record; unowned records vanish from everyone's view in most systems.

Keep the raw export untouched in a separate folder. If a cleaning step goes wrong, you restart from the export, not from a half-cleaned file.

Field and pipeline mapping

Every source field needs a destination or a decision to drop it, written in a table the whole team can read. Pipelines are where meaning shifts: the old stage called Proposal might mean sent, or accepted, or under negotiation, and the new system needs one of those.

Test import and checks

Never run the full import first. Take a sample of fifty records that includes the awkward cases, a contact at two companies, a deal with no value, a note with an attachment, and import it into an empty sandbox or the real system with a tag you can bulk-delete. Then check by hand:

1. Counts: the same number of contacts, companies and deals in as out, minus what you deliberately dropped.
2. Links: open a contact and confirm the company, deals, notes and owner are attached.
3. Dates: created dates and activity dates match the source, not the import day.
4. Pipeline: deals sit in the right stage and the total value per stage matches the old report.
5. Search and filters: the fields the sales team uses daily behave as expected.
6. Automations: nothing fired during import, no welcome emails or task floods to migrated contacts.

Fix the mapping, delete the sample, repeat until a full pass is clean. Two or three rounds is normal. This is the step that turns cutover from a gamble into a rerun of something that already worked.

Cutover day and freeze

Choose a day with low sales activity, tell the team a week ahead, and freeze the old system at a set time: read-only, or simply an instruction that nothing is entered after that hour. Take a final export, import it with the tested mapping, run the same checks, and only then switch integrations, website forms and email connections to the new system. Keep the old system in read-only for a few months as a reference, then close it and delete the data in line with your GDPR retention rules.

When you do not need help with this: a small team moving a few hundred contacts with no history can do it with the vendor's import tool in a day, following the checks above. Where it justifies a CRM implementation is volume, years of history, several integrations, or a team that cannot afford a week of confusion, and the cost lines for that are in our CRM implementation cost guide. If you are unsure which, a short call with an export open is enough to tell.

Frequently asked questions