Problem

The Fortune 500 auto parts manufacturer wanted to reduce waste and the number of parts returned post purchase by distributors and customers.
Parts were sold in kits of product bundles and the company had more than 10,000 parts with more than 100 different product bundles. Which product bundles would minimize waste and maximize customer satisfaction?

Process

Data gathering for this company required weeks of cleaning and preparation. With many missing values it became important to fill in those values manually by verifying them with other internal sources. A good predictive engine is a result of good data and therefore data prep is the primary and one of the key steps in data modelling.
After successful validation and consolidation of relevant data, extensive statistical analyses and knowledge sharing with business was done to remove variables that are irrelevant in predicting purchase propensities of different product bundles.

After many iterations, hundreds of cross-selling opportunities and perfect product bundles were identified by the AI engine that were fine tuned to reduce waste and post purchase returns by customers. Moreover, proven sales and marketing strategies were proposed to win these customers.
The sales teams were able to increase their client win rates by 3x and reduce their sales research and analysis hours by 60% on average - this included product research hours, information search and finding the next action step to close the deal.