Map meaning as well as names
| Source | Target | Transformation | Reject or review | Owner |
|---|---|---|---|---|
| cust_code: 00042 | external_id | Keep as text, including leading zeros | Duplicate source identity | Data owner |
| status: A | active | A → true; I → false | Any code outside A/I | Operations |
| joined: 03/04/2026 | joined_date | Confirm source date convention first | Ambiguous until confirmed | Source system owner |
| name: empty | display_name | No invented default name | Missing required value | Record reviewer |
Add the source type, target type and one valid and invalid sample to your working copy. A blank, zero and “unknown” may have different meanings. Decide those meanings before conversion. Keep the untouched source export separately so rejected rows can be traced.
Check totals without hiding rejected records
For a synthetic batch of 120 rows, 113 accepted plus 7 rejected accounts for every input row. It does not prove the 113 values are correct. Compare selected identities, inactive statuses and date boundaries against the source; record corrections as a new mapping revision before repeating the pilot.
A numeric export turns 00042 into 42. Can the import simply add two zeros to every ID?
Show the worked answer
Only if the source identity format is confirmed and collision checks pass. Otherwise obtain an intact text export or an authoritative identity mapping; a guessed repair can merge different records.