The same product may have a different title, SKU or pack description on HKTVmall, Shopify and Shopee HK, and building the mapping by hand, one product at a time, slows down onboarding.
Why match products first?
Matching links each listing on a platform to the physical product in the warehouse. Only once matching is done can the stock, price and sales data for the same product on different platforms be organised on the same basis.
- Product: the physical item in the warehouse, identified by Barcode.
- Listing: how a product is sold on a particular platform, including its title, price and images.
- Matching: linking a listing to the product it belongs to.
How does exact Barcode matching work?
When a platform listing's Barcode is the same as the product's Barcode, the two are matched directly. A Barcode match is exact and its result is unambiguous, which is why it is the first step in matching.
When there is no Barcode, how does AI fill the gaps?
When a platform listing has no Barcode, or the Barcodes do not agree, RAID's AI product matching proposes likely matches based on title similarity. It covers the products that exact Barcode matching cannot reach and helps build the cross-platform product mapping.
What if a match is wrong?
Every AI match is presented as a suggestion and takes effect only after a person approves it, so a wrong match is never applied automatically. In the review screen, the team can confirm, correct or reject suggestions one by one; the system never changes product data without approval. Listings that could not be matched can also be matched by hand.
Is AI matching also used for competitor products?
Yes. Merchants choose their own competitors; the AI matches competitor products to yours and marks them 'pending confirmation' for checking, then uses your sales volumes to estimate how competitor price cuts affect your revenue. Unmatched or out-of-stock SKUs are re-matched automatically on a schedule.
How does RAID handle this?
RAID's cross-platform product link layer connects listings, SKUs, Barcodes and Bundles, tying platform products to physical goods. Once matching is done, product names, descriptions, specifications, images, prices and BBD (best-before date) can be organised in one pass, then loaded to the chosen stores and platforms.
Key points
- Exact Barcode matching comes first; AI title matching handles cases with no Barcode.
- The 91% figure for AI title matching is an internal benchmark that depends on product data quality.
- Every AI match suggestion must be approved by a person before it is applied; AI competitor matches are marked 'pending confirmation'.
- Matching is the foundation for bringing stock, price and sales data together across platforms.