Reviewing a card

Status: pilot. Draft — needs review before publishing.

Review is the point of the product: you decide on the few fields the system is not sure about instead of building the card from scratch.

Confirmed facts and proposed values

Every attribute carries its own status, and where possible the source it came from:

• values confirmed against a source, with a link and the snippet they were taken from;
• values the system proposed from the rest of the data.

Both can be right. The difference tells you where to spend your attention.

The quality summary

Each card carries a summary:

Field	What it tells you
Completeness	How much of the structure expected for the category is filled
Confidence	How reliable the filled values are
Validation status	The overall state of the card
Missing attributes	Attributes expected for the category that are still absent
Low-confidence attributes	The fields that most need your decision

Completeness is only meaningful once the expected attribute set for a category is known. Until then, read it as an indicator, not as a guarantee.

What can be corrected

• the product fields — name, SKU, brand, category;
• the content — title, descriptions, bullets, FAQ, SEO fields, image alt texts;
• the identity — brand, model, MPN, category, canonical title;
• the attributes — value and unit.

A corrected attribute is locked, so a later automatic pass cannot overwrite your correction. During the pilot, corrections can be sent back by the same channel the card arrived in.

One thing to keep in mind

Product identity and facts are shared per barcode across the sellers who use it. Correcting the brand, the model or the category improves the record for everyone — and changes what other sellers see as well. Content, prices and stock levels stay with you.