The Opportunity
- A consumer goods company spent several years and millions of dollars building a demand signal repository – a collection of store-level point-of-sale data from multiple retail customers.
- They desired to transition from simple status reporting to forward-looking recommendations, through predictive analytics, that they could share with their retail business partners.
Our Approach
- In our strategy explorations we identified consumer segmentation as the top first strategy to pursue. Consumer packaged goods companies do not sell directly to consumers, but operate through their retail customer channels. So consumer-oriented segmentation can be more challenging.
- With access to the retailer’s store level data, we theorized that consumers could vary in their price responsiveness at each store. Traditional store-based segmentation systems, including those offered by vendors, use attributes, rather than behavior, to group stores. In our approach, we use actual consumer behavior (price elasticity) as the basis upon which to group consumers.
The Impact
- Among the retailers selected, we estimated thousands of price elasticities, for each item-store, for products in three of the client’s categories. Those elasticities were then used to feed statistical clustering routines, to find distinctive groupings of stores.
- On average we found that about six groups or types of stores existed, each with distinct profiles of how shoppers would respond to pricing. Further profiling revealed differences in total spending and market basket combinations.
- The creation of store groups allowed the retailer and the company, in conjunction, to more precisely target consumers with a more tailored pricing strategy.