Clean a contact list before CRM import: a step-by-step checklist

An import is the one moment you can fix your data cheaply. After it, every duplicate becomes two half-histories, every badly formatted mobile becomes a failed SMS, and every contact with unknown consent becomes a risk each time someone sends a marketing message. Clean first.

This matters even more in an agent-native CRM, where agents read and write records all day. An agent working from a clean list keeps it clean. An agent working from a messy one writes more mess, faster. The checklist below works whether you're moving from spreadsheets, an email account's contacts, or another CRM.

Step 1: Gather every source and freeze them

Export everything into one working folder before you change anything: the old CRM, accounting contacts, email contacts, booking system, spreadsheets on people's laptops. Keep the raw exports untouched and work on copies. Tell the team that the old sources are frozen from a date; new contacts after that go into a holding sheet until the import is done.

Add a column to every file called 'source', filled with where the row came from. You'll need it when two sources disagree.

Step 2: Agree the columns you'll keep

Map every export to one standard set of columns. Anything that doesn't map either gets a deliberate home or gets dropped. A sensible small-business set:

Keep postcodes as text, not numbers, or spreadsheets will strip the leading zero from Northern Territory postcodes such as 0800.

Step 3: Fix Australian phone formats

Phone numbers cause more failed imports than anything else. You'll find 0491570159, 491 570 159, (02) 5550 4321, +61 (0) 491... and numbers with notes like 'after 3pm' typed in. Pick one storage format and convert everything to it.

The Australian Government Style Manual gives the national formats as 02 5550 4321 for landlines and 0491 570 159 for mobiles, and the international format as +61 2 5550 4321 and +61 491 570 159, dropping the leading zero (Style Manual: Telephone numbers). Store numbers in the international form: it's unambiguous for SMS and calling systems, and you can always display the national form.

• Strip spaces, brackets, dashes and '(0)'.
• Ten digits starting 04: mobile. Replace the leading 0 with +61.
• Nine digits starting 4: almost certainly a mobile missing its 0. Add +61.
• Ten digits starting 02, 03, 07 or 08: landline. Replace the leading 0 with +61.
• 13, 1300 and 1800 numbers: keep them in their national form, in their own field; they don't fit the landline or mobile pattern.
• Anything else, or text in the field: move it to a 'phone notes' column for a person to check.

Mobiles and landlines go in separate columns. If you'll ever send SMS, you need to know which is which.

Step 4: Merge duplicates with clear rules

Deduplication is where people either give up or merge too aggressively. Use three tiers:

1. Exact email match: treat as the same person. Merge automatically.
2. Same normalised mobile, different email: probably the same person with two addresses. Merge, keeping both emails, after a quick check.
3. Similar name and same organisation, nothing else matching: possible duplicate. Don't merge automatically; list them for a person to decide.

When merging, keep the most recent value for each field and note which source it came from. Never merge on name alone; two different Michael Nguyens at two different businesses are a real thing.

Step 5: Record consent, or mark it unknown

If you'll send marketing email or SMS, consent has to be on the record. The ACMA explains that there are two types, express and inferred, that inferred consent is less reliable than express, and that under the Spam Act it's up to you to prove you have consent, so you should keep a record of who gave it, when and how (ACMA: Avoid sending spam).

In practice, most old lists don't have that evidence. Don't invent it. Use four values: express (with date and method), inferred (with the relationship that supports it), none or unknown, and opted out. Anyone marked opted out in any source stays opted out after the merge. This is general information, not legal advice; if marketing is a big part of your business, get advice on your specific list.

Step 6: Remove what you don't need

An import is a good time to delete. Old leads with no contact in years, suppliers you no longer use, fields like date of birth that nobody uses for anything. If you're covered by the Privacy Act 1988, APP 11 includes obligations to destroy or de-identify personal information in certain circumstances, and APP 10 asks for reasonable steps to keep what you use accurate and up to date (OAIC: APP quick reference). Smaller lists are also simply easier for agents and people to keep current.

Step 7: Test with 20 rows, then import

Pick 20 rows that include your awkward cases: a merged duplicate, an international number, a 13 number, an opted-out contact, an NT postcode, a name with an apostrophe. Import those first. Open each record in the new CRM and check every field. Fix the mapping, delete the test records, then run the full import. Spend the first week after import reviewing a sample of records.

Switching CRM without losing the history

Contacts are the easy part of a migration. The context is harder: notes, past emails, deal history, the reasons behind lost deals. Before you switch:

• Export notes and activities, not just contacts. Check the export actually includes them.
• Map old deal stages to new ones in a table, and decide what happens to deals in stages that don't exist any more.
• Bring across open next actions so nothing due next week disappears.
• Keep a read-only copy of the old system, or its full export, for at least as long as you might need to look something up.

If you're weighing a move away from a larger platform, our HubSpot alternative comparison for Australian businesses covers the trade-offs. Once the clean list is in, agents keep it that way by following the logging rules in agents that write the CRM for you. New researched leads should go through the same checks; see AI prospect research.

The one-page checklist to clean a contact list

1. Export every source; keep raw copies; add a source column.
2. Freeze old sources from an agreed date.
3. Map to one standard column set.
4. Normalise emails, names, addresses and postcodes.
5. Convert phones to +61 format; separate mobiles from landlines.
6. Merge exact-email matches; review phone matches; hand-check name matches.
7. Record consent as express, inferred, none or opted out; carry every opt-out across.
8. Delete what you don't need.
9. Test import 20 tricky rows; fix the mapping.
10. Import, then sample-review for a week.

Fulcrum CRM is designed so agents keep contacts, deals and notes current after the import. See how it works on the CRM page, and check which systems connect on integrations.