Want to know more?
Are you interested in this project? Or do you have one just like it? Get in touch. We’d love to tell you more about it.
How automated product bundling increased average order value by 5% at johnlewis.com
Transforming personalised basket recommendations at John Lewis & Partners
Most online retailers suggest additional products to customers when they add an item to their basket. This method of increasing sales is often done by suggesting products previous customers have also bought when ordering the item they originally searched for. John Lewis, the iconic British retailer, had previously taken a manual approach to this strategy, but the data team wanted to find a more efficient way to use data from purchasing behaviour online to increase spend through related, personalised product recommendations.
As a long term partner of John Lewis & Partners (JL&P), Equal Experts got involved in the development of an automated product bundling feature to drive increased sales and conversion. The potential generated by initial trial results has since led to this personalisation strategy being adopted across multiple departments in the business. Here’s how collaboration, innovation, and a relentless focus on improving the customer experience has led to an expected annual increase in sales of 5%.
Outcomes
5%
Add to cart increase
5
departments now using the model
8
months uplift consistently sustained since launch
About John Lewis & Partners
John Lewis & Partners is one of the UK’s oldest, largest and most popular retailers. They operate 34 stores across the UK, as well as johnlewis.com. It has total trading sales of £4.93 billion, a workforce of 38,000 Partners (employees), and is part of the John Lewis Partnership – the largest employee-owned business in the UK.
Industry
Retail
Organisation size
38,000 partners
Location
UK
Project length
5 years
Challenge
Increasing revenue and improving customer experience more efficiently
At the heart of this work was a task that had been tried before but with limited success. Previously, the team had manually curated related product recommendations to suggest to customers when they added specific items to their baskets (e.g. a TV stand when buying a TV). However this approach was cumbersome and didn’t scale well with the store’s constantly changing inventory. Particularly in departments like tech and beauty, where new products are launched frequently, it was difficult to keep up to date with manually curated recommendations.
In the world of personalised product recommendations, the key was relevance – ensuring that recommendations not only matched the customers’ search terms, but also anticipated what they might want next. Previous attempts to automate the process had relied on hand-written SQL rules, which – although they gave trade teams good information about past customer purchasing behaviour to inform appropriate product recommendations – were far too slow to keep pace with the dynamic nature of the inventory. JL&P needed a more automated, flexible solution.
Solution
Automating product bundling recommendations with machine learning
The goal was to increase revenue with personalised recommendations via an automated system that used machine learning to respond to rapidly changing product availability and adapt to customer behavior. The new system would have to be smart enough to recommend products that made sense together and complemented what was already in the client’s basket – TVs paired with HDMI cables, for example, or skincare products based on previous beauty purchases.
Building on the client’s existing infrastructure using Google Vertex, the team developed a sophisticated machine learning-powered recommendation engine that could automatically surface relevant complementary products based on previous customers’ browsing and purchase behavior, removing out of stock items or products less likely to convert, leaving only those with a high likelihood of appealing to the customer and being added alongside the original purchase.
As well as past purchase patterns, the new feature also factors in attributes like colour to make even more relevant suggestions in departments like home and fashion. Using existing technology allowed us to create more tailored, flexible personalisation models better suited to the unique needs of the business.
At the same time, the team also wanted to make the new recommendation carousels far more engaging, with better graphics, cleaner layouts, and more intuitive navigation, to improve the entire customer journey. The goal was to give the customer as much information as possible on the page, to enable them to make the right decision for them. We also wanted to make the recommendations feel like an integral part of the shopping experience—not just something tacked on at the end of the page.
Result
5% sales uplift from recommended products added to customer carts
An initial test during the pre-Christmas sales peak saw a 5% increase in sales value per customer session, a significant gain in a competitive retail landscape. This uplift in average order value has remained consistent over the subsequent 8 months of operation, and the bundling capability has been expanded across more website departments, including Home, Garden, Electricals as well as some areas within the Baby & Child and Sports & Leisure departments.
The automated nature of it means the recommendations are always up-to-date, and the data-driven approach makes the suggestions much more compelling for customers. By leveraging existing Google Vertex AI models, we were able to analyse past customer interactions and make more accurate predictions about what customers would be likely to respond to. This deep dive into customer behavior data has enabled a level of personalisation that hadn’t been possible before.
Lead Consultant
, Equal Experts
Conclusion
The team is now exploring ways to further enhance the bundling experience, including potentially surfacing recommendations more prominently on product pages. A/B testing of the bundling feature is planned for John Lewis apps within the first year. “There’s definitely more we can do to make it an even more integral part of the shopping journey,” concluded Simon.