MRO · Master Data · Lâm Văn Phong
MRO cataloging: clean item data without creating procurement risk
Two similar descriptions do not necessarily identify interchangeable parts. Good cataloging lets buyers, storekeepers and engineers identify the right item while preserving transaction history.
01
Distinguish items, spares and rotating assets
The Item Master describes a type of material; a rotating asset tracks an individual serialized instance. Before merging codes, find every storeroom, BOM, work order, contract and purchase history that uses each code.
02
Standardize descriptions
Apply noun–modifier rules by commodity class and define required attributes such as size, material, rating, manufacturer part number and unit of measure. Preserve the original description and provenance. Do not force every commodity into one attribute template.
03
Find duplicates by confidence
Separate exact duplicates, likely duplicates and possible equivalents. Rules and AI can suggest candidates; an engineer must verify fit, form, function and warranty conditions before a merge. Record why two codes stay separate when manufacturer or service conditions differ.
04
Migrate with controls
Build an old-to-new crosswalk and check stock, open purchase orders, reservations, BOMs and accounting values. Pilot one commodity class, reconcile record counts and review rejected rows. Retire old codes only after transactions and users have moved.
05
Govern the clean catalog
Define an item-request form, commodity approvers, duplicate checks and metrics for code-creation time, missing attributes and inactive items. Without an owner, the catalog will degrade again.
06
Design a commodity-specific taxonomy
A standard description must say what the material is and which parameters determine substitution. A bearing may require type, bore, outside diameter, width, clearance, and seal; a gasket may require material, size, pressure class, and applicable specification. Define attributes by commodity with units, allowed values, and a way to mark unverified data. A manufacturer part number is valuable identification, but it is not a full technical specification.
Retain three layers: raw description for traceability, normalized description for search, and approved technical specification for purchase or substitution. AI may extract attributes, but store confidence and the source text. For high-consequence parts, an authorized engineer should approve the record; data-cleaning staff alone should not decide technical equivalence.
07
Decide duplicates and interchangeability
An exact duplicate is truly the same item under the same service conditions; a possible duplicate is only a review candidate. Interchangeability also depends on fit, form, function, rating, material, certification, warranty, and duty. Two part numbers may be equivalent in one machine but not another. An alternative-part relationship therefore needs an effective scope by asset class or BOM, an approver, and an effective date.
Prioritize review by stock value, purchase frequency, lead time, and equipment criticality. Do not use the number of deleted codes as an isolated success measure; a wrong merge can cause an incorrect purchase, lost traceability, or a technical nonconformance. Keep evidence for decisions to merge, retain separate codes, or create an approved alternative relationship.
08
Migrate and govern the Item Master
Before cutover, build an old-to-new crosswalk and assess stock balances, open purchase orders, reservations, BOMs, vendor catalogs, contracts, and transaction history. Do not delete an old code if it is needed for historical lookup; use inactive or superseded status where the EAM supports it. Test units and conversions too: the same part may be bought by box and issued by piece.
New-item requests should carry commodity, minimum attributes, an image or datasheet when needed, a duplicate check, and an approver. Measure creation time, returned requests, new duplicates, and use of approved alternatives. Review exceptions monthly rather than waiting for another cleansing project. Good cataloging reduces purchasing errors and makes the right part easier to find; it does more than shrink the code count.
09
Illustrative example: two similar bearing codes
Suppose two bearing codes share dimensions and near-identical descriptions, but differ in clearance and seal. Duplicate detection flags a possible duplicate rather than merging it. An engineer reviews datasheets, BOMs, and duty at each installed position; stores checks stock and open orders. If substitution is approved for only one position, the Item Master retains both codes and records an alternative relationship scoped to that BOM, with approver and effective date. If they are truly the same item, the migration plan maps old codes, updates reservations and purchase orders, and retires the old code through the agreed process. Reporting should show reduced wrong-purchase exposure and duplicate checks on new requests, not just a lower code count.
Implementation checklist
- Commodity taxonomy and required attributes
- Old-to-new mapping and source records
- Technical equivalence approval
- BOM, stock and open-transaction reconciliation
- Data owner and new-item rules
Practice framework by Lâm Văn Phong. Check the standards, rules and current data applicable to each project before use.