What is catalog intelligence?
Catalog intelligence is the use of AI to turn raw product and listing data into structured, reviewed product records: enriched attributes, mapped categories, variant and offer groups, and duplicate and policy decisions.
In plain words: Raw data arrives as supplier files and seller listings, each with its own titles, categories and gaps. Catalog intelligence reads that data and proposes a structured record: attributes filled where the evidence supports them, a category from your tree, variants and offers grouped, duplicates found and policy issues flagged. People review the proposals before they are used.
Published by Youzu · Editorial policy
- Published
Diagram
3 sources become one reviewed record
Why it matters
A marketplace or retailer often receives the same product from several sources, described differently each time. Without a structured record, filters miss products, duplicates split reviews and stock across listings, and policy problems reach shoppers. Catalog intelligence moves that clean-up earlier, before records reach the PIM, the storefront or a channel.
How it works
Read
Supplier feeds, seller listings and product images are read together.
Propose
Attributes, a category, variant and offer groups, and duplicate and policy findings are proposed.
Review
Reviewers accept, edit or reject proposals, and uncertain cases wait in a queue.
| Field | Seller's original | Proposed |
|---|---|---|
| Title | jkt olive | Olive jacket |
| Color | OLV | Olive |
| Style | Empty | Utility |
| Material | Empty | Left empty: sent to review |
| Category | Empty | Fashion › Jackets |
Examples
One messy supplier row
A row titled 'jkt olive' with the color 'OLV', no material and the description 'wool' becomes a reviewed record: a plain title, an allowed color value and a category leaf, with the material conflict sent to review.
Duplicates across sellers
6 sofa listings from different sellers resolve to 2 products with their offers, while a 2-seat version of one sofa stays separate as a variant of the same family.
Policy before publishing
An image rule against watermarks, applied to the first image of 6 listings, flags 3 of them for review before they reach search.
Common mistakes
Treating proposals as approved data. The PIM or marketplace platform stays authoritative; reviewed records go there, not the other way round.
Judging it on a demo set. A clean demo hides the messy records. Test on your own sample, catch-all categories and near duplicates included.
Related terms
- Product data enrichment
- Adding missing information or improving existing information on a product record so products can be found, filtered and compared.
- PIM (product information management)
- A central system to store, enrich, govern and distribute product information to sales channels.
- Asynchronous ingestion
- Asynchronous ingestion accepts product data first and processes it afterwards, so an accepted request is not yet a finished product.
In Youzu
Youzu helps marketplaces, classifieds and retailers turn raw seller and supplier product data into structured, reviewable catalog records.
Sources
Product data specification, Google Merchant Center Help. support.google.com/merchants/answer/7052112. Accessed .
Global Trade Item Number (GTIN), GS1. gs1.org/standards/id-keys/gtin. Accessed .
FAQ
Questions buyers ask
No. A PIM stores, governs and distributes product information your team has already approved. Catalog intelligence works earlier in the flow: it reads raw feeds, listings and images and proposes the structured record that then goes into a PIM, search or marketplace. Many teams use both, with the PIM staying authoritative for its own data.
It should not. A sound setup proposes values and decisions, and people or your own rules decide what is applied. 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.
Use a sample of your own catalog, including the difficult records, and judge the output against labels your team trusts. Measure accuracy on your own sample against your own labels, with critical errors reported separately. Evaluations start with a sample of your own catalog and expand to more workflows once the results are reviewed.
Last reviewed
Related
Keep reading
Product
Catalog Intelligence for marketplaces and retailers
Catalog Intelligence maps products to your category tree, extracts category attributes, keeps variants and offers apart and flags issues for review.
Glossary
What is product data enrichment?
Product data enrichment defined in plain words, with a worked example, common mistakes and links to the guides and Youzu products that apply it.
Glossary
What is a product information management (PIM) system?
A PIM defined in plain words, with a worked example, common mistakes and links to the guides and Youzu products that apply it.