The order
Both are hard to change once thousands of parts reference them. See Parts.
Every part that will be referenced by anything, including sub-assembly parents.
Chosen by somebody who did not prepare the file.
Parent, child, quantity. Every referenced part must already exist.
Last, because they reference parts.
So the imported state is recorded before anybody edits it.
Import
structures-2026-03.csv · 316 rows · validating
| Row | Parent | Child | Qty | Check |
|---|---|---|---|---|
| 1 | NW-1200 | 204118 | 1 | OK |
| 2 | NW-1200 | NW-1210 | 1 | OK |
| 3 | NW-1210 | 100248 | 14 | OK |
| 4 | NW-1210 | 180999 | 1 | Unknown part |
| 5 | NW-1300 | 311902 | 0 | Quantity is zero |
Cleaning before loading
The best time to remove duplicates from a part master is before it is a part master. Once structures reference both copies of a duplicated part, merging is a rewrite rather than a deletion.
- Sort by description and read it. Duplicates are obvious to a person and invisible to a diff.
- Normalise units and case before loading, not after.
- Decide what to do with parts that are used in nothing. Usually: do not load them.
- Load a subset first — one product, end to end — and look at it before loading everything.
A completed import cannot be undone in one action. A structure import that references parts that do not exist stops partway, leaving some structures loaded and some not — which is recoverable but tedious. This is why validation runs first and why loading one product before loading four thousand is worth the extra hour.
What this does not do
Undoing an import means deleting what it created.
Import is a load, not a connection to another system.
There is nowhere for documents to go.
The module holds neither.
Questions
Can you do the import for us?
Yes, as part of an implementation engagement.
What format?
Tabular. The expected columns are shown on the import screen.
Can we re-import to update?
Re-importing creates rather than updates. Ask before attempting it on a live part master.