Clean up the database The duplicate queue,
settled by a decision
When a hunch is not enough, the pair waits for a human. Four steps, no side spreadsheet.
- 1
The data quality band
At the top of the customer list: how many have a document, an email and people with access, how many possible duplicates exist and how many people arrived with no company attached. Every number filters the list.
- 2
The reason, and how likely it is
Each pair comes with the why — same tax number, same email, same phone, very similar name, same CNPJ root for a branch office — and a Very likely or Likely mark.
- 3
Side by side, and you decide
For customers the merge goes field by field: you choose which side each one comes from, anything empty on the survivor is filled in from the one leaving, both sets of emails and phones stay, and to confirm you type the name of the record that goes. For people you choose who stays and the rest follows.
- 4
Not a duplicate
One click takes the pair out of the queue for good. Keeping two records costs far less than merging the records of two different people.
Outside the queue: people who arrived with no company attached get a customer suggestion — from the customer contact list, from their tickets or from their corporate email domain, never from gmail and the like — and the consistency check lists what is out of place, with a one-click fix whenever the fix is safe.