Most item data problems are not created in the ERP. They are created the moment a new item is set up: a supplier spreadsheet is copied in, a description is typed from memory, a unit of measure is guessed. By the time anyone notices, the error has already travelled into purchasing, inventory and pricing.
Fixing a wrong item afterwards means finding every order, stock balance and report it has touched. Stopping it at the door means checking one row. That difference is why the cheapest place to improve master data quality is before the data enters the ERP or PDM, not after.
“An item master is only as good as the moment each item was created.”
What goes wrong at item setup
When a new product or spare part is added, the person doing it is usually under time pressure and working from whatever the supplier sent. Four questions tend to go unanswered:
- Does this item already exist under another code or name?
- Is the classification the one purchasing and reporting actually use?
- Are units, pack sizes and dimensions consistent with similar items?
- Who is responsible for the item once it is live?
Why fixing it later costs more
Once an item is in use, a correction is no longer a data change. It becomes a process change: open orders, stock counts, price lists and integrations all carry the old value. Teams learn to work around errors instead of fixing them, and the workarounds become part of how things are done.
Duplicates split the picture
Two codes for the same item mean two stock levels, two purchase histories and two prices. Neither tells the whole story.
Missing attributes block decisions
Without material, origin or dimensions, the item cannot be compared, substituted or sourced from another supplier.
Inconsistent classification hides spend
If similar items sit in different categories, spend analysis and supplier negotiations start from the wrong numbers.
A check before the system, not a new system
Item data can be checked before it is written into the ERP or PDM, without replacing either. Incoming rows are compared against existing items, completed from reliable sources and flagged when something looks off. The item master stays where it is; what changes is what gets into it.
| Step | What is checked | Who decides |
|---|---|---|
| Match | Possible duplicates and existing equivalents | Item owner confirms |
| Enrich | Missing attributes, with the source shown | Reviewer accepts or edits |
| Classify | Category against the agreed structure | Category owner |
| Release | Completeness against required fields | Written to ERP or PDM |
Where to start
Start with a sample rather than the whole item master. A few hundred rows from one product group are enough to see which errors are common, which rules are missing and how much of the work can be automated.
- Which item group causes the most questions in purchasing?
- Which fields are most often left empty?
- Who already corrects items by hand, and how?
Action plan
- Pick one item group. Choose the one where errors cost the most.
- Run a sample check. Measure duplicates, missing fields and classification gaps.
- Agree the rules. Decide required fields and who owns each category.
- Check new items at the door. Apply the rules before items reach the ERP or PDM.
- Extend group by group. Use what was learned to cover the rest of the item master.
- Most item data errors are created at setup, not in the ERP.
- A correction after go-live is a process change, not a data change.
- Checking items before they enter the system needs no new ERP or PDM.
- Start from one item group and a sample, then extend.