Fix the cause, once
When a load shows a problem, IDN's root-cause analysis (RCA) looks for the cause on its own, shows what it blocks and who owns it, and proposes the fix. The fix is made once, where the cause is, and the next load checks it again, so the same defect does not return in every cycle.
| The project way | Teams fix the record in front of them; the error returns next cycle. |
| The IDN way | RCA finds the cause, proposes the fix, and the next load confirms it. |
The problem
Most teams fix the failed records in front of them. The cause stays, so the same error comes back in the next cycle, and finding out why a record failed takes days of back-and-forth between teams.
How IDN does it
- Finds the cause: RCA traces a failure back to where it starts.
- Shows impact and ownership: what the cause blocks and who owns it.
- Proposes the fix: the fix is made once, at the cause.
- Confirms it: the next load checks it again.
From our own testing
A customer master data load wrote 1,603 of the 56,679 customer company-code records it should have. RCA found the cause on its own: dunning data, which most customers never have, had been set up as required, so every customer without it was dropped. RCA proposed one change, making that data optional. After the change and one re-run, all 56,679 records loaded.
What it means for the program
Fewer repeat defects, shorter cycles, and less time lost to investigation.
Questions
How do you find the root cause of data migration load failures?
Trace the failed records back through the rule, mapping and source data until you reach the single setting or record that causes them. IDN's RCA does this automatically and shows what the cause blocks and who owns it.
Why do the same defects return in every mock cycle?
Because teams fix the failed records instead of the cause. Unless the cause is fixed at the source, the next extraction brings the same defect back.
What is root-cause analysis (RCA) in data migration?
A method for finding the underlying reason a group of records failed, such as a mapping rule, a configuration setting or a source data issue, so it can be fixed once.
Next: 6. Source to sign-off →