A leading US based office products retailer wanted to measure and optimize ROI of items featured in its weekly circular. It had so far been using a control store for this purpose but now required robust modeling to minimize sales loss and determine effectiveness. The data that was available was in disparate databases and formats.

• A 4 member AbsolutData team worked in close tandem with the client for 5 months. Additionally, one member spent time onsite with the client team to understand the requirements better.
• The AbsolutData team developed a robust model using Hierarchical Bayes Estimation to predict Store level base line sales at sub category level. This included seasonality, store level characteristics and tied into Annual Sales Forecast developed by the client.
• We had to apply automation technique to bring disparate databases together.
• We automated the model execution. The weekend’s report was available by Tuesday morning.
• The report included incremental sales/margin, Halo/Cannibalization impact by store, category, business, region and buyers.

• ROI was increased by about 20%.
• Led to changes in the mix of items featured in Weekly Circular.
• We were asked to roll out similar model for the client’s Canada business.