Case Study: Dixons Carphone boosts product coverage and add-to-basket rates with Syntasa

Dixons Carphone leverages behavioral data with award-winning use of AI to improve customer experience

Dixons Carphone, a leading multinational consumer electrical retailer, faced the challenge of manually creating product bundles for its online store, a process that lacked personalization and was difficult to scale. Their existing data from Adobe Analytics was not readily usable for analysis, and they lacked the extensive data engineering resources needed to build and productionize a machine learning recommendation system at scale, especially to handle highly seasonal traffic spikes. They engaged vendor Syntasa to address these challenges.

Syntasa implemented a solution that integrated natively within Dixons Carphone's GCP environment to synthesize behavioral data. They built and productionized a personalized recommendation engine using a Nearest Neighbor model for customers with sufficient history, alongside a non-personalized Natural Attach algorithm. As a result, product coverage (the share of product views showing a recommendation) doubled from 32% to 72%, and personalized AI-driven recommendations led to a 3x increase in add-to-basket rates.

Dixons Carphone

Paula Bobbett

Head of Online Performance