Find and merge duplicate records

Nobody creates a duplicate organization on purpose. It just happens. A rep searches for "Hartley Engineering", finds nothing and creates it. A colleague had already added it a year ago as "Hartley Engineering Ltd". A lead comes in through the form with a personal email address, while the same person is already in the CRM with their work address. And after an import from an old system, some organizations appear twice, once with and once without the town in the name.

It looks like a small problem. Until an account manager calls a customer who was called by a colleague last week, because the activity sat on the other record. Or until the sales manager pulls revenue per customer and sees the biggest account split in two. Or until someone unsubscribes from mailings and still gets a campaign through the second record.

Duplicates make your timeline incomplete, your reports unreliable and your consent uncertain. So you do not just want to clean them up once. You want to keep finding them. That does not work by hand. You cannot compare thousands of persons with each other, and when you merge you have to decide which phone numbers, which owner and which consent survive.

In Lumo the Duplicates screen does the searching. Rules decide what counts as a duplicate and each pair gets a confidence level. You decide on the doubtful ones. Our guide on CRM data quality explains why clean data is the base for everything else in a CRM.

Say a technical distributor with 15 account managers moves to Lumo. In the old system some organizations appeared two or three times, because each account manager created their own customers. Here is an example of how a clean-up round could go.

After the import the admin runs Search now for organizations. Pairs come out with different confidence levels. Before merging anything, he looks at the rules. Many organizations have "Supplies" or "Ltd" in the name, so he adds those to terms that never count. A large customer's main number appears on forty persons, and the shared number limit stops those persons from being matched with each other.

Then he searches again. The Certain pairs, often the same domain and a nearly identical name, he merges with Merge all certain. The Likely pairs he goes through with the account managers in an hour, because they know whether "Baker Installations North" and "Baker Installations" are the same customer. For Review pairs he often picks Not a duplicate, because they are separate branches he would rather record as parent and child organizations.

Next he turns to the persons queue. A week later he checks the Log. One merge turned out wrong, so he undoes it. The rest stays, and new duplicates created in the weeks after appear in To review on their own.

Duplicates mostly appear at the point of entry. You manage persons and organizations in contact management, where you can also select several records in a list and merge them. If you come from Pipedrive, the import is idempotent: a second import run does not create new duplicates of what is already there. See switching from Pipedrive.

To stop the team creating incomplete records, use the data quality rules in CRM adoption and guidance. After a clean-up, totals per customer in reports and dashboards are right as well.

A merge is not the same as erasing a person on request. If someone has to leave your CRM completely, use the GDPR tools in security and privacy. Blocked addresses there make sure an erased person does not come back through a form or an import.