Hypermarket Case Study
- SEGMENT: Grocery
- PRODUCT: RichRelevance Find™, Recommend™, Discover™ and Engage™
- CHALLENGE: Grow digital revenues by enhancing customer experience and engagement on e-commerce and mobile app. Help customers build the basket faster with AI-driven advanced merchandising, reducing dependence on manual curation by a merchandiser. Provide contextual search results and personalize web banners for higher engagement.
- RESULTS: The supermarket chain implemented the full suite of personalization platform to create individual experiences throughout the site and achieved
19.3% Items Per Order
+3.92% Average Order Value
6.15% CTR on Recommendations
*Results from Recommend™ only
A lifestyle holding company in the Middle East has exclusive franchisee of this renowned French retailer, operating over 150 hypermarkets and supermarkets in 30+ countries, and serving over 200,000 customers a day. Their online business includes e-commerce and mobile app, supporting both English and Arabic, and RichRelevance deployment spans 8 countries.
In 2019, the retailer was looking to improve their commerce experience through meaningful personalization through recommendations, search and content. Engagement was low as a result of generic banners, and no recommendation engine. Through its physical stores, a lot of purchase data was available, but it was not used for online personalization. Search functionality had a different problem, where a keyword search (such as mobile phone) resulted in huge laundry list of results and a frustrating process for the shopper, trying to find what they need.
To make the shift to individualized recommendations, contextual search and better product discovery, RichRelevance started a phase wise deployment of the solution suite.
Starting with the ‘cold start’ scenario, i.e., when a first time or unidentified shopper visits, recommendations based on products trending in the geo-location are immediately shown.
For example, a person adds eggs to the basket, and cross-sell recommendations include products that people in Dubai’s Palm Jumeirah typically buy with eggs (milk, potatoes, chicken or cheese), helping to quickly build the basket. Additionally, the UX is tuned to grocery, where people buy from category page itself. The easy ‘add to cart’ buttons on cross-sell recommendations ensure conversion.
Another application of personalization is to use intent signals to identify affinities and promote relevant products in real-time. As an example, a shopper browses the ‘fresh food’ menu and arrives at ‘fish’ category (through Discover™), then proceeds to add ‘organic sea bass’ to the cart. Recommend™ algorithms detects the preference for ‘organic’ products, and suggests other organic products across produce and poultry.
Other personalization applications include personalized search, where a shopper with purchase history is shows very targeted products for the same search term, compared to an anonymous shopper. When a customer who has purchased an iPhone in the past, and has a Macbook in the cart searches for ‘mobile phone’, sees variants of iPhone 11 upfront, while an anonymous buyer sees a variety of phones trending in the city. This leverages deep catalog coverage, where offline purchase data is also considered to recommend products online.
Other applications include use of advanced merchandising strategies to ‘complete-the-recipe’, where machine learning is leveraged to serve cross-category recommendations based on cart components. As a shopper searches for, and adds pizza base to the cart, she sees related products on the same page – pizza sauce, cheese and soda, dramatically increasing click through and improving basket value.
The retailer has also recently rolled out content personalization through Engage™, to personalize categories shown on home page, while retaining merchandiser control to boost or bury categories based on business goals.
Initial results show an exceptional click-thru rate of 6.15% on product recommendations and an uplift of 19.28% on units per order. Furthermore, customers can now build their baskets much quicker, reducing friction and findability issues associated with a vast catalog.