Variant vs Duplicate vs Replacement in Product Data

A duplicate is one product under two records, a variant is a different product sharing a parent, a replacement is a successor. Telling them apart is an identity decision, not a mapping step.

A duplicate is the same product under two records. A variant is a genuinely different product that shares a parent, such as the 10A and the 16A breaker. A replacement is a successor meant to supersede another product. Telling them apart is an identity decision, not a mapping step, and each misread corrupts a different downstream process.

Supplier onboarding is where the three get confused, because a new feed arrives with part codes that do not match yours and nothing in the file says which relationship applies. Claro treats each as a separate recorded relationship, so the call is made once, with evidence, and reused on the next feed.

The worked example: one breaker, three codes

A distributor’s ERP holds a 10A single-pole breaker. A new supplier feed arrives:

Supplier code SB-1P-10 with the description “MCB 1P 10A C curve”.
The manufacturer’s own reference, matching the ERP’s manufacturer part number.
A sibling, SB-1P-16: same family, same description pattern, 16A rated current.

SB-1P-10 and the ERP row are a duplicate once the manufacturer reference matches: one product, two records. SB-1P-16 is a variant: it shares the parent but is a different product, so it must stay separate. If the manufacturer later discontinues the 10A part and names a successor, that link is a replacement.

Relationship	Meaning	Correct handling	If misread
Duplicate	Same product, two records	Merge into one canonical record	Inventory and pricing split across records
Variant	Different product, shared parent	Keep separate, link to the family	Merged as a duplicate: the wrong part ends up in a quote
Replacement	Successor intended to supersede	Link old to new, flag the old as superseded	Missed: discontinued stock gets ordered

Why this is an identity decision

Description similarity cannot separate these cases: the 10A and 16A breakers differ by one character. The deciding evidence is the manufacturer reference, the rated-current attribute and the datasheet. A decision to treat two references as one product is stored as a SAME_AS decision; the opposite call is stored as NOT_SAME_AS, so the pair is not re-merged on the next run. This is the core of entity resolution.

Glossary

Entity resolution

How records are resolved to real-world products.

Glossary

SAME_AS decision

The recorded judgment that two references are one product.

Playbook

Group product variants into one family

Structure variants so search and inventory make sense.

Playbook

Cold-start supplier onboarding

Seed a new supplier's feed with what you already know.

To check a supplier file for repeats, try the Duplicate SKU Finder.

FAQ

What is the difference between a duplicate and a variant?

A duplicate is the same product held under two records. A variant is a genuinely different product that shares a parent, such as the 10A and 16A version of the same breaker. Merging variants as duplicates puts the wrong part in a quote.

What is a replacement in product data?

A replacement is a successor product intended to supersede an older one, usually after a manufacturer discontinues a part. It is a relationship between two different products, not a duplicate and not a variant.

Why is this an identity decision and not a mapping step?

Mapping moves a value from one field to another. Deciding whether two records are the same product, siblings or successors requires evidence about the product itself, so it is an identity decision that should be recorded with its evidence.

What does each confusion cost?

Duplicates split inventory and pricing across records. A variant treated as a duplicate puts the wrong part in a quote. A missed replacement leads to ordering discontinued stock.