Delivering On-Demand Operational Insight
Built a market basket intelligence layer that revealed which products customers naturally purchase together across millions of transactions.
The Challenge
A global retailer operating thousands of locations across North America lacked visibility into how customers actually built their baskets. Leadership could analyze item-level sales, but transaction data and product data lived in separate systems. As a result, the organization could not answer critical merchandising questions: • Which products customers naturally purchase together • Which items drive basket expansion • Which combinations create the strongest promotional opportunities Several internal efforts attempted to build a self-service reporting solution, but the scale of the data made the problem difficult. Millions of transactions across thousands of products created performance constraints that prevented the delivery of a tool that was both fast and usable. Merchandising and marketing teams were ultimately making bundling and promotional decisions without visibility into real purchasing behavior.
Our Approach
Rather than attempting to analyze the entire product catalog, we reframed the problem around business impact instead of data completeness. Working with data engineering and business leadership, we identified the few hundred highest-impact products across the North American business. These products represented: • <20% of the product catalog • >70% of total company revenue Focusing on these products dramatically reduced query complexity while still capturing the transactions that drove the majority of business performance. On top of this foundation, we built a market basket intelligence layer that enabled product affinity analysis across transactions at scale. This shifted the organization beyond traditional attach-rate metrics toward true product affinity, revealing which products customers were most likely to purchase together within the same order. The solution delivered: • A consolidated executive view of high-impact product relationships • A detailed transaction-level exploration layer for deeper analysis
The Result
Leadership gained the organization’s first clear view of true customer basket behavior across thousands of stores. Teams were able to: • Identify high-affinity product combinations • Design more effective promotional bundles • Improve cross-sell strategies around high-traffic products • Prioritize merchandising decisions based on real purchasing patterns What had previously been an unsolved technical challenge became a scalable decision engine for merchandising and promotional strategy.