How RecoMelon handles new products with no sales history

What the cold-start problem is

Recommendation engines that depend on purchase history naturally favour established products. A newly launched SKU has few or no behavioural signals, so it can remain invisible precisely when the merchant wants customers to discover it.

How ColdStart™ works

ColdStart™ uses product imagery and VisualDNA™ to establish useful relationships between a new product and the existing catalogue before transactional history develops.

What RecoMelon can use immediately

• Visual similarity to existing products.
• Image-derived attributes from catalogue intelligence.
• Category and availability context.

Where new products can appear

Once processed, eligible products can enter Shop Similar and other RecoMelon discovery experiences rather than waiting to become statistically popular.

What merchants need to do

Keep product records active, provide representative product imagery and ensure the product is available in the relevant market. Large catalogue changes may require processing time before every relationship is available.

ColdStart™ versus manual merchandising

Merchants can still promote a launch deliberately. ColdStart™ solves a different problem: it gives the recommendation system a product-level understanding of the new SKU even before customer behaviour exists.

How to validate new-product discovery

Check representative new arrivals against related products, confirm they appear in appropriate recommendation contexts and monitor engagement as live shopper data begins to accumulate.