Personalised recommendations lift online sales for an electronics retailer

Btech is one of Egypt’s leading electronics and home appliance retailers and among the Middle East’s largest authorised resellers of brands such as Apple, Samsung, Bosch, Ariston and LG. Customers buy through more than 80 stores and an online store, with consumer financing available on many products.

After its acquisition by UK-based private equity firm DPI, Btech set out to grow online sales with a larger online range and a better shopping experience. With more than 7,000 products and a fast-growing base of over 15 million repeat customers, shoppers needed help finding the right items quickly, and the store needed a way to show each customer what was most relevant to them.

Working closely with Btech’s sales and marketing teams, I designed and delivered a recommendation engine built on the store’s own purchase and click data and on the manufacturers’ product specifications. It combines three methods:

• Personal recommendations. A collaborative filtering model learns from purchase histories which products customers with similar behaviour buy, and produces the top ten picks for each customer.
• Similar products. A content-based model compares products on their technical attributes (size, capacity, power, features such as Wi-Fi or no-frost) to suggest close alternatives on every product page.
• Smarter category pages. Category listings rank products by current promotions, including discounts and interest-free instalments, and then by sales over the last 30 days.

The pipeline refreshes daily from the online store’s order and product data and publishes recommendations for the website to display.

• Online sales up by more than 15% within three months of deployment.
• Recommendations cover the full catalogue of more than 7,000 products and update every day as new orders arrive.