Helping millions find their style

Like all good personal stylists, you show the customer what you know they like. And then a little something to push the limits on their style.

  • banner

    We didn't just give customers what worked.
    We gave them what matched.

  • banner

  • banner

Details, Numbers, and Highlights

The Challenge

Macys aimed to improve the recommendations of items a customer might be interested in while browsing and/or making a purchase on They targeted three key areas of improvement:

1. Improve the relevancy of offers (of clothing or other catalog items) to the customer, based on the CURRENT browsing experience, gauging what the customer is interested in based on items in the customer's basket or items they have browed.

2. Drive incremental revenue

3. Improve the customer's online experience; open doors to new styles that would interest the customer just like personal stylist would

The challenges included:

  • Millions of visitors on site; with tight constraints on any increased latency to allow for decisioning
  • And a large product catalog, with thousands of items; making it difficult to predictively model each individual item

The Solution

The answer to how to increase relevancy of offers while driving incremental revenue was locked in a large amount of data. Our approach was to systematically understand customer purchases of catalogue items, considering all factors including time of year/season, individual buying characteristics, demographics, regional preferences, and a lot more. Next we sought to bring order to an understandably large product catalog. We did this by creating "facets" of goods that complement each other. It is against these facets that we developed our predictive modelling solution.

The Results

An overhauled recommendation engine capability internally at Macys, and tied into the complete product catalog. The uses are limitless, from straightforward 'Add to Cart' offers, to emerging uses such as online personal shopper, or location based mobile offers...