What is a false merge?
A false merge joins 2 records that describe different products; a missed match leaves 2 records of the same product apart.
In plain words: Matching can go wrong in 2 ways. A false merge joins 2 different products into one record. A missed match leaves 2 records of the same product apart. The first hides a real difference; the second only repeats a product, so teams usually report the 2 apart.
Published by Youzu · Editorial policy
- Published
Why it matters
A false merge is hard to see and hard to undo: reviews, stock and prices of 2 products blend into one. A missed match is visible as a duplicate and easy to fix.
How it works
Compare
Check identifiers first, then model, key attributes and photos for each pair of records.
Decide
Merge only when the evidence supports it; otherwise keep the records apart.
Measure
Count false merges and missed matches on a labelled sample and report them separately.
| Merged | Kept separate | |
|---|---|---|
| Same product | Correct merge | Missed match |
| Different products | False merge | Correct separation |
| Cost if wrong | Hidden difference, hard to undo | Visible duplicate, easy to fix |
Examples
2 sofa sizes
The 2-seat and the 3-seat navy sofa share a model name.
A matching code, a different size
2 jacket listings carry the same code but sizes M and L. Treat the clash as a conflict to review, not as a match.
Two sellers, one product
Two sellers list the same armchair with the same code and photo.
Common mistakes
One error rate for both. A single accuracy figure hides whether the method merges too much or too little.
Code equals match. A shared identifier is strong evidence, but a different size or model beside it is a conflict to review.
Related terms
- Product matching
- Deciding whether 2 records describe the same product by comparing identifiers, titles, attributes and images.
- Product variant
- One purchasable version of a product that differs along defined dimensions, such as size, color or pack size.
- Critical error
- A critical error is a mistake an evaluation reports on its own line because one instance matters, such as a wrong allergen or a false merge.
In Youzu
Youzu helps marketplaces, classifieds and retailers turn raw seller and supplier product data into structured, reviewable catalog records.
Sources
Global Trade Item Number (GTIN), GS1. gs1.org/standards/id-keys/gtin. Accessed .
Data Matching: Concepts and Techniques for Record Linkage, Entity Resolution, and Duplicate Detection, Springer. link.springer.com/book/10.1007/978-3-642-31164-2. Accessed .
FAQ
Questions buyers ask
Usually the false merge, because it hides a real difference and is hard to undo, while a missed match leaves a visible duplicate. Your own costs decide. Whichever you choose, count the 2 errors apart and report both on the same labelled sample.
Matching is precision-first: when 2 listings cannot be shown to be the same product they stay separate, and uncertain cases can go to review. A pair that is left apart can be merged later by a person, but a wrong merge can lose the difference it hid.
No. Products, variants and seller offers are separate records: the same product sold by several sellers is one product with several offers. A variant stays its own record under its product, so sizes and colors of one jacket are grouped but not collapsed into a single record.
Last reviewed
Related
Keep reading
Product
Duplicate detection and product matching
Precision-first duplicate matching: listings stay separate unless they can be shown to be the same product, and uncertain cases can go to review.
Glossary
What is product matching?
Product matching defined in plain words, with a worked example, common mistakes and links to the guides and Youzu products that apply it.
Guide
Product, variant, family and offer: a data model for marketplaces
A clear data model for marketplaces and retailers: how a product family, a variant and an offer differ and why keeping them separate matters.