Which duplicate should you keep?
Deleting duplicates is easy; deleting the right ones is what matters. If two rows share an email but one has a phone number and company and the other doesn't, you almost certainly want to keep the more complete row.
| Option | Keeps | Use it when |
|---|---|---|
| First | The earliest occurrence | The file is sorted oldest → newest and the original record is the reference. |
| Last | The latest occurrence | Later rows are updates (e.g. an export appended to an older one). |
| Most complete | The row with the fewest empty cells | Rows came from different sources with different fields filled in. |
Choosing the right key
The key is the column (or columns) that identifies a real-world record:
- Contacts and leads: email address — the identifier most CRMs match on.
- Customers, orders, invoices: the customer ID, order ID or invoice number.
- Products: SKU.
- No reliable ID? Use a composite key such as first name + last name + phone.
Rows where the key is empty are never treated as duplicates of each other — ten contacts with no email are not the same person.
Normalise first, then deduplicate
Exact comparison misses duplicates that differ only in formatting. “Ravi Kumar ” (with a trailing space) and “Ravi Kumar” are different to a strict comparison, as are “ASHA@X.COM” and “asha@x.com”. TheSpreadFix ignores case and extra spaces when comparing, and when you apply email or phone fixes together with deduplication, the values are cleaned first so the duplicates are caught.