How VisualDNA™ Works

ChromaSense™ interprets colour from product imagery so RecoMelon can distinguish shades, tonal families and compatible colours beyond inconsistent catalogue colour names. It can support colour-led matching, refinement and merchandising across large catalogues.

IntentWeave™ reads product imagery and converts visible characteristics into structured catalogue signals such as silhouette, neckline, sleeve, texture, pattern and colour family, reducing the need for manual tagging and helping new products enter discovery faster.

VisualIndex™ is the catalogue intelligence layer that makes visual product relationships usable across RecoMelon discovery. It organises product imagery and attributes so Shop Similar, search, collections and other experiences can retrieve relevant products without relying only on merchant tags or purchase history.

RecoMelon match quality is based on granular visual and catalogue signals rather than a single similarity score. VisualDNA™ considers relationships such as silhouette, colour, pattern and structure, while Dealbreaker™ Filters and shopper behaviour can further refine which alternatives are useful in context.

RecoMelon differs from conventional recommendation engines by starting with product and visual intent rather than requiring historical co-purchase data. Its visual catalogue intelligence supports new products immediately and powers shopper-controlled refinement, conversational search, complementary sets and journey recovery.

InsightCore™ connects recommendation impressions and clicks with deeper discovery behaviour and influenced commercial outcomes, helping teams understand which RecoMelon surfaces, products and shopper decisions contribute to conversion and AOV.

ColdStart™ uses VisualDNA™ and product imagery to give new and low-history products meaningful recommendation relationships before sales data accumulates, so new drops can enter Shop Similar and wider discovery from day one.

VisualDNA™ analyzes product imagery to match items by visual similarity, not purchase history. This helps Shopify brands show more relevant recommendations from day one, including new products.