MandM Direct Drives 2.4% Revenue Increase with Personalized Real-Time Recommendations

Online discount apparel retailer MandM Direct, which had previously relied on standard rules-based recommendations to guide website visitors through the shopper journey, has generated impressive improvements in multiple KPIs with real-time, personalized recommendations.

Results from delivering AI-powered deep learning Google Recommendations through Google Cloud AI and via the Qubit platform include:

  • 2.4% increase in overall revenue;
  • 114% jump in overall product recommendation click-through rates (CTRs);
  • 0.7% conversion rate (CR) uplift from targeted recommendations on the site’s home page; and
  • 157% increase in CTRs from recommendations incorporated into add-to-cart pop-ups on product detail pages.

“We hadn’t used a huge amount of ‘deep learning’ prior to this deployment, although we did use a bit of machine learning on the next purchase preference,” said Paul Allen, Head of Ecommerce at MandM Direct in an interview with Retail TouchPoints. “We knew the theory behind it was right, but getting the right message in the right place was a challenge.”

MandM Direct had been a Qubit customer since 2013 and welcomed the opportunity to serve as a proof of concept for the Google Recommendations solution, which was deployed in April 2020 after tests earlier that year. The recommendation engine determines not just what a particular website visitor might like to see, but also the optimal points in the shopper journey to place these recommendations. Location options include the site’s home page, product landing pages (PLPs), product detail pages (PDPs), pop-ups when an item is added to the cart and the checkout page itself.


Allen explained how a customized site visit shaped by deep learning might proceed. “If they start on our home page, we would put specific recommendations there to show visitors the range of products they might be interested in,” he said. What’s shown will be different for each visitor, “based not just on the shopper’s on-site behavior but how they got to this page,” Allen added. “Things like the browser they used and the type of device can indicate what this customer is most likely to respond to.”

If the visitor then goes to a category page, for example for sneakers (known as “trainers” in the UK), the solution will add “adidas products if it thinks you’re an adidas shopper, but it will also show another brand as well,” Allen said. “If the visitor then goes to another brand, say Skechers, the solution would determine that the person is trying to buy trainers, but they are probably brand-agnostic about which one they buy. The recommendations are constantly updating based on browser and visitor behaviors.

The solution even addresses one of the most common customer complaints about online shopping: being shown the same product that the consumer has just bought. That’s why for placements on an add-to-cart or checkout page, “the solution ‘knows’ the customer has adidas trainers in the basket, so it’s not trying to sell you another pair of trainers,” noted Allen.

Even where the recommendations appear on the page is a potential variable. “We’ve seen that placing recommendations on the PLP adds a lot of value even when it’s on the bottom of the page,” said Allen. “We want to test whether we can put it further up on the PLP. Our overall goal is to get relevant products in front of the customer, wherever they’re likely to see them. This shortens their customer journey and, hopefully, they buy more from us.”

Retailer Gains Valuable Insights About Products and Customers

In addition to the KPI improvements from the solution, MandM Direct has seen related benefits. “We’ve seen a wider range of products being recommended, because it’s taking on more data points,” said Allen. “With a rules-based engine, you tended to get more top sellers, but you can see in the data that more people are seeing that wider range.”

Some of the solution’s biggest successes have been in multiplying CTRs. “The increases in click-through rates show just how relevant the product recommendations are — the products are what the customer wants to see,” said Allen.

The retailer can make adjustments that shape the recommendations, and MandM Direct is constantly testing the impact of these changes. “You can set the goal as, for example, raising conversion rates or increasing average basket size,” said Allen. “Then you can test whether increasing the add-to-basket goal drives more revenue than boosting the conversion goal.”

Allen and his team at MandM Direct are continuing to explore the potentialities of the solution by expanding the types of customer data sources it draws from with additional in-session data. The goal is to define the customer’s “mission” each time they visit as a way to provide even more targeted personalization.

Featured Event

Join 5k+ attendees for an immersive experience where you can get inspired by retail’s most innovative minds and see their big ideas in action.



Access The Media Kit


Access Our Editorial Calendar

If you are downloading this on behalf of a client, please provide the company name and website information below: