Developed customized recommendation engine for a leading e-commerce firm, increasing cross-sell rates by 60%

Marketing, Digital Businesses

We drove 60% increase in cross-sell rates for a leading e-commerce client. To achieve this, we developed a real-time customized recommendation engine that predicted the most optimal array of cross-sell products to be shown to a visitor. The engine was based on the contents of the shopping cart, visit behavior, & other parameters, and took into account the probability of a cross-sell as well as the value of a successful cross-sell.

Approach

  • e-commerce client wanting to increase cross-sell rates.
  • Identified orders with multiple sales and appropriately tagged cross-sell and bundled products.
  • Developed exhaustive list of parameters to be used to predict likelihood of cross-sell.
  • Used probability prediction modeling to identify high propensity cross-sell products depending upon these parameters

Analysis

s3

Results

  • Designed an optimal matrix of cross-sell products based on expected value of cross-sell.
  • Client implemented the matrix across the platform.
  • Evaluated the results of the pilot; Updated matrix led to 60% increase in cross-sell propensity.
  • Refined matrix further based on results of the pilot