Store Clustering

Brief

We helped asports goods retailer, who was operating 150+ stores in analysing the store variables to implement storeclustering for better store management

Challenge

Our client wanted to do away with uniform strategy and adopt personalised strategies across stores to have competitive advantage. For the same, the store clustering was implemented However, it was trick to choose the variables to cluster the store and equally difficult was to choose the right clustering techniques considering the client’s business objective.

Our Action

• Collected the fragmented store related data including sales revenue, revenue growth, local pricing, store size, Store geography information, age of the store, product split ,customer demographics, local competition- along with macro factors to understand the data distribution and key statistics.
• Determine the variables those are most relevant in dividing your stores into clusters considering the client’s  business problem
• The shortlisted variables were reduced to a relatively smaller set and  excess variables were removed  using  Principal component analysis (PCA)
•  We then developed multiple cluster models with multiple iterations; in the end combined  theselected relevant models to reach  to the clustering ensemble.

Results

• With cluster based inventory planning, overall  inventory cost reduced
• Improvement in  Customer’s Net Promoter Score with reduction in detectors
• 10% reduction in stock outs