CRM Data Migration Checklist: 7 Phases From Audit to Cutover

Switching CRMs, or moving off a spreadsheet, rarely fails because the new software is bad. It fails because 4,000 contacts arrived with 600 duplicates, every deal shows today's date as its creation date, and half the notes are attached to nobody. The team stops trusting the data in week two and drifts back to the inbox.

This CRM data migration checklist prevents that. It runs in seven phases, each with a short list of steps you can tick off, plus a field-mapping example, the failures we see most often, and realistic timelines by record count. It applies whether you are leaving another CRM or a spreadsheet; our CRM vs spreadsheet guide covers when to make that move.

What a CRM migration actually moves

A CRM holds more than a contact list, and each type of record moves differently. Contacts and companies are easy: one row each, imported first. Deals are harder, because each one must link to a contact or company and land in a stage that exists in the new system. Activities and notes are harder still; most import tools want them in a separate file linked by email address or an external ID. Users and record owners must exist in the new CRM before anything is assigned to them.

Some things usually do not move at all. Full email threads rarely import; you keep the old system for reference or forward the important ones. File attachments often need manual re-uploading. Audit history, who changed what and when, is almost never portable. Decide up front which of these you can live without, and write it down so nobody is surprised at cutover.

Check the destination's limits before you start. HubSpot's free plan caps at 1,000 contacts and Bigin's free plan at 500 records. monday CRM Basic allows 1,000 contacts and Standard 10,000. If your export has 3,000 rows, you need a paid tier before the import will complete; our best free CRM guide lists every cap.

Phase 1: Audit what you have

You cannot clean what you have not counted. This phase takes an hour for a spreadsheet and half a day for a CRM with several years of history.

1. Export every object raw. Contacts, companies, deals, activities, notes, and users, each as its own CSV. Save these originals in a folder you never edit.
2. Count records per object. Write the numbers down. These are the totals you will reconcile against after the full import.
3. List every field and how full it is. A column that is 4% populated is a candidate to drop. A custom field nobody remembers creating is a candidate to ask about.
4. Identify the records that matter most. Open deals, active customers, and anything touched in the last 12 months. These get spot-checked by hand later.

Phase 2: Clean the data in a spreadsheet

Fix problems in a copy of the export, not after import. Editing 2,000 rows in a spreadsheet takes an afternoon; editing 2,000 CRM records one by one takes a week.

1. De-duplicate contacts. Match on lowercase email first, then on name plus company. Merge, do not delete, so notes from both copies survive.
2. Standardise formats. Dates to one format (yyyy-mm-dd is safest), phones to one pattern with country code, currency values as plain numbers without symbols.
3. Remove dead records. Leads with no activity in two years and no email address are noise. Archive them in a separate file rather than importing them.
4. Fix picklist values by hand. If your old stage column contains "Proposal", "proposal sent" and "Prop", pick one and replace the others now.

Phase 3: Map every field

Import wizards auto-match columns whose names look similar. They get "Email" right and quietly skip or misfile "Stage", "Owner" and "Created". Build the map yourself, in a table like the one below, before you open the wizard.

1. Create custom fields and picklist values in the new CRM first. The wizard can only map to fields that already exist.
2. Create every user before importing deals. Owners that do not match a user get dumped on the importer's account.
3. Write the picklist map explicitly. Old value on the left, new value on the right, for stages, sources, types and statuses.
4. Decide what happens to unmapped columns. Drop them, or park them in a "Legacy notes" text field. Do not let the wizard decide.

Phase 4: Import a test batch

Twenty-five records is enough to find every mapping error and small enough to delete cleanly if it goes wrong. Pick records that use every field, including the messy ones.

1. Import 25 contacts and companies, then 10 linked deals. Use real records from your cleaned file, not the first 25 rows.
2. Open five records and compare them field by field to the source. Check dates, owners, stages and phone formatting specifically.
3. Fix the mapping, not the records. If a stage landed wrong on one record, it landed wrong on all of them. Correct the map and re-import.
4. Delete the test batch completely before the full import. Otherwise the full run creates duplicates of your test records.

Phase 5: Run the full import

Order matters. Import parents before children so links resolve: companies, then contacts, then deals, then activities and notes. Run it outside working hours if the team is still using the old system.

1. Freeze edits in the old system. Announce a cut-off time. Anything entered after it will need re-entering by hand.
2. Import in dependency order. Companies, contacts, deals, then notes and activities, checking the wizard's error report after each file.
3. Save every error file. Rows the wizard rejects are the ones you will fix and re-import, so keep them separate from the originals.

Phase 6: Verify before anyone uses it

Verification is where you catch the errors that did not throw an error. Budget half a day for it.

1. Reconcile counts. Records per object in the new CRM should equal your audit totals minus the duplicates and dead records you removed on purpose.
2. Spot-check 20 records across objects. Include your most important open deals. Compare dates, owners, stages and amounts to the source.
3. Check the three usual casualties. Deal stages, creation dates and record owners. Filter for deals created today, deals in the default stage, and records owned by the importer.
4. Confirm notes landed on the right records. Open five contacts with long histories and read the notes. Orphaned notes are the hardest thing to fix later.

Phase 7: Cut over and keep the old system read-only

Cutover is a decision, not a technical step. From a named day, the new CRM is the only place work gets recorded.

1. Pick a Monday. Import over the weekend, verify on Sunday, and start everyone on the new tool on Monday morning with a 30-minute walkthrough.
2. Make the old system read-only. Remove edit rights or cancel down to a free or view-only plan, but do not delete it.
3. Keep it for 30–60 days. Something will surface that did not migrate. Knowing you can look it up removes most of the anxiety.
4. Export the old system one final time before it closes. Store the CSVs somewhere permanent, even if you never open them.

The rest of the launch, from user set-up to first-month adoption, is covered in our CRM implementation checklist.

What breaks most often, and how long it really takes

How long the whole thing takes depends on record count and on whether you are moving deal history and notes, not just contacts. The estimates below assume one person doing the work part-time alongside their normal job.

If your migration is in the bottom two rows, ask the new vendor what migration help is included. Salesforce Pro Suite and HubSpot Professional (which carries a $1,500 onboarding fee) both come with structured onboarding that covers import; entry tiers generally leave you with the wizard and the help centre.

Frequently asked questions