E-commerce Personalisation and Recommendation Engines
Personalisation — adapting the shopping experience to individual customers based on their behaviour, preferences, and attributes — is one of the highest-ROI investments in e-commerce. Amazon attributes 35% of its revenue to its recommendation engine. Netflix credits recommendations for $1B+ in annual value retained. For most retailers, personalisation remains an underexploited opportunity.
Personalisation Dimensions
- Product recommendations: "You may also like," "Frequently bought together," "Recently viewed," "Customers who bought X also bought Y"
- Homepage and category curation: Different users see different featured products based on their preferences and browsing history
- Search personalisation: Search results ranked by personal relevance, not just text match
- Email personalisation: Product recommendations in email based on browsing and purchase history
- Pricing personalisation: Targeted promotions and discounts for specific customer segments
Recommendation Approaches
- Collaborative filtering: "Customers like you also bought..." — finds similar users and recommends what they bought. Requires sufficient user data.
- Content-based filtering: Recommends similar items to ones viewed/purchased — based on product attributes. Works with less data.
- Hybrid: Combines both approaches — better results than either alone
Implementation Options
Specialist personalisation engines: Nosto, Dynamic Yield (acquired by Mastercard), Barilliance, Monetate. Integrated with platforms: Shopify — LimeSpot, Frequently Bought Together. Enterprise: Adobe Target, Salesforce Personalization. Build vs buy decision depends on catalogue size, traffic volume, and engineering capability.