Top

Would you like to?

TALK TO SALES JOIN OUR TEAM

Advanced analytics helps leading US automotive parts retailer improve merchandising and store performance

  • Client:

    Leading US automotive parts retailer

  • Industry:

    Retail

Download a copy

Business challenge

A leading US automotive parts retailer was facing a major challenge with variable store performance. In particular, stock supply and demand seemed to be misaligned – both across individual stores and across different categories of products.

As a result, between 25% and 70% of demand was getting filled by purchasing stock from external vendors. This sales pattern was having a huge negative impact on the retailer’s profit margins.

The company asked The Smart Cube to help improve the operations of its 2,000 stores across the US.

The Smart Cube solution

The Smart Cube used predictive analytics and optimisation algorithms to respond to the client’s business challenges.

Exploratory analysis was conducted to evaluate patterns in inventory, sales, consumer demographics, and market attributes data. The scope of this exercise included ~100,000 area-SKU combinations for two major product categories. The team tested multiple statistical techniques, such as ARIMA, random forests, and regression, to evaluate best-fit, and narrowed down on hierarchical mixed-effects models to predict product sales and identify key drivers of sales at each store/area.

Further, optimisation algorithms were deployed to align product assortment at each store in-line with forecasted demand, thereby highlighting opportunities in inventory allocation for each store.

An interactive dashboard was created to enable users to generate monthly forecasts and to identify optimal product assortment at store level, based on expected demand. 

Results

  • Potential savings by stocking the right products: The analysis revealed potential savings of over $11 million per year, by stocking the right products to increase sales, and also to reduce reliance on products purchased from external vendors.
  • Prevented over-stocking of products: The output of the optimised forecast revealed the client was over-stocking some products by ten times the sales. This helped identify the SKUs that should be taken off the shelves.
  • Helped in understanding factors affecting sales: The model estimates helped the client understand the importance and extent to which various factors impacted the sales of a product. The estimates were generated at each level, the most granular level being SKU.

Value delivered

Understanding the factors affecting sales and optimising inventory led to tremendous benefits across multiple teams.

Purchasing team

  • Effective planning in procurement of products
  • Efficient use of resources
  • Better allocation of funds by effective sourcing and distribution of products

Merchandising team

  • Effective and efficient use of shelf space
  • Prevent under-stocking and over-stocking of products
  • Better coordination with suppliers

Central Analytics/BI team

  • Better understanding of seasonality/trend in the data
  • Answer core business questions by understanding of important variables in determining product sales