Identical product codes do not always identify the same item. Equally, one item may have different codes in sales software, inventory records and a supplier's file. Connecting these sources without agreeing their meaning can deduct the wrong variant or combine quantities measured in different units.
Product-code preparation is therefore more than tidying names. It creates a traceable relationship showing what a code from a particular source means, when that relationship applies and who confirmed it.
Source codes and central references have different roles
Preserve the original code together with its source system. The same characters may identify different products in different applications. Treat the source and code together as the reference rather than assuming the code is unique throughout the business.
A central product reference allows confirmed matches to share a history. It does not require every application to change its codes immediately. A mapping can retain familiar operational references while connecting them to an agreed product record.
Names are useful for reading but unreliable as the only joining key. Spelling, abbreviations and word order vary. Similarity matching can suggest candidates for review, but should not confirm a match while the variant or unit remains uncertain.
Preserve leading zeros and original formatting
A code with leading zeros can change meaning when a program treats it as a number. Store codes as text and apply explicit rules to spaces and invisible characters. Removing all punctuation to make codes look alike can merge values intended to be distinct.
Keep both the original value and the normalised comparison value. If a supplier changes its file format, this makes the transformation traceable instead of forcing someone to reconstruct it from the final result.
Distinguish the product, variant and selling unit
Different colours, part sizes or capacities may require separate inventory even when they share a product name. Check the attributes that affect selling and fulfilment before combining records.
Units deserve the same attention. Items sold individually, in packs or in cartons need confirmed conversion rules. Do not add quantities from two systems without knowing what each unit represents, or assume that one conversion factor fits every package.
GS1 describes the GTIN as an identifier for trade items that are priced, ordered or invoiced. Such a standard identifier can help identify a product, but is not the same thing as every business's internal code and does not automatically resolve unit mappings.
Product bundles need explicit relationships
A bundle sold under one code may consume several inventory items. Mapping it to a single item without its components loses that meaning. Define what is consumed by one bundle and the period for which the composition applies.
A different situation is several codes acting as aliases for the same product. Keep alias relationships separate from bundle relationships. Treating both as an identical kind of match can distort quantities and values.
Give the mapping a responsible owner
Useful mapping fields include source system, original code, central product, variant, unit, conversion rule, review state and effective date. Unresolved records need a review queue with reasons, such as missing size or unknown pack quantity.
The data owner should confirm matches using operational evidence, such as a supplier catalogue or inventory record. Choosing whichever item looks closest merely to make an import succeed sends the uncertainty into orders and reports.
When a mapping changes, retain the editor, reason and effective period. Decide whether it applies only to new transactions or whether historical records require a controlled correction. Discontinued codes should remain traceable and should not silently be reused for a new product.
Assign responsibility for incoming codes as well as the initial cleanup. Otherwise the mapping begins to deteriorate when a new supplier, sales channel or product range is added. The process should make uncertainty visible rather than relying on someone noticing an unusual total later.
Rehearse cases likely to reveal a wrong match
Test ordinary products, similarly named variants, leading-zero codes, different selling units, bundles and returns. Follow each case from order to stock movement and reporting, checking that the quantities retain the same meaning.
Define whether an unknown code stops processing or enters a review queue. Avoid automatically assigning it to a generic product. Specify who corrects the reference and how held transactions resume afterwards.
After launch, track unresolved mappings and repeated causes. If system stock still differs from physical stock, investigate movements alongside product references. For businesses receiving orders through chat, the chat-to-sales-order guide explains how confirmed products and units can move into the next operational step.
Connected systems, but product codes still need manual matching every day?
Talvyn Labs can review product records, define code mappings and align units so sales, inventory and reporting share traceable data.
Prepare product data for integration