A large retail organization in the organic food specialties market wanted to improve its online ecommerce transaction amount per customer. The ecommerce site offered product catalog, shopping cart, order confirmation, credit card processing, and account management but did not have an effective recommendation engine for cross-sell and up-sell.
Data & Analysis
The team met with stakeholders and developed engagement goals and objectives. We then analyzed customer purchasing behavior and patterns online as well as offline (in-store environment). The analysis suggested a three-step approach. First step was to identify distinct clusters within the customer base, the next step was to understand behavior and purchasing patterns for each segment, and the final step was to develop the recommendation model.
The team utilized unsupervised learning model to identify distinct customer clusters and their respective attributes. For each cluster, we utilized machine learning and association rules to develop the recommendation engine model. The model takes into account the real behavior and buying patterns of customers.
The recommendation engine is a powerful tool that is not only informing the in-store product placement strategy but also their online cross-sell placements. The engine continues to learn from the new buying and behavioral patterns and incorporates these learning into the model on an on-going basis.