To optimise our client’s stores in China to grow store revenue per square meter.
By applying a Retail analytics solution, we were able to incorporate internal and external data to drive the following outcomes:
By leveraging on machine learning algorithms, our solution provides clustering for more than 300 stores based on similarity of customer preferences.
The optimal number of segments were selected based on a silhouette score which assesses the similarity a store is to its own cluster (cohesion) in comparison with other clusters (separation).
Each segment was then given a scorecard which helped group these stores based on customer preferences such as preferred product categories, average spending power etc.
The following table is an illustrative example of the store segmentation exercise:
* The illustration above has been simplified and does not contain all the information from the segmentation exercise.
As you can see in Segment 1, the customers were categorised as Trend Adopters, who preferred skinny fit jeans for bottoms and blazers and tees for tops. These customers were not that interested in premium collection tapered jeans, shirts and sleeveless tops but at the same time they were also not price sensitive.
Segment 2 seems more oriented towards Premium collection tapered jeans for bottoms, shirts for men’s tops and sleeveless tops for women. Whereas segment 3 were categorised as price sensitive due to their preferred basic collection straight fit jeans, hoods and graphic tees.
Following this, we mapped these stores which belonged to the respective segments across China as shown below:
* The above diagram is just for illustrative purposes and does not contain the entire segment information.
This provided the opportunity for our client to apply customised segment-level execution strategies pertaining to promotions planning, pricing, markdown and effective product assortment.