Retail stores behave like Real Estate.
There is competition between brands to have their products on specific shelves and close to high valuable products. As a retailer, it is important to understand the value of each location, based on the position on the shelf, nearby products and customers' demand.
But there are metrics that are not being totally explored on this highly dynamic environment, such as:
In this analysis, we will be using data from a retail brand that has physical facilities in several countries. The sales data shows 541909 transactions captured between 01/12/2010 and 09/12/2011 (1 year).
Before we go deep using AssetFloow's API, let's take a look at the raw data:
The sales amount is not equally distributed for all countries, but we will take leverage on that in this analysis:
There is one store on each country. Since the amount of sales for the United Kingdom is way bigger, that store will be defined as ideal. We will use the patterns of that store as the base to find anomalies in sales on the other stores' location, in this case, in France.
First, we can use AssetFloow to visually understand how each product brings value to different parts of the store. We drew a retail store layout, with shelves represented in dark blue, seen from above. Let's use the API function SpatialMap() on the sales data from United Kingdom in two products, Product 15 and Product 27, and understand their spatial influence in the whole store. The heatmaps below show the correlation between that specific product and all others on the store (more transparency means lower correlation).
The reason we are showing these two maps is because these two products show what is the diference between an essential product (such as fish, meat, bread, milk), with strong correlation with the majority of products of the store, and a product that is non-essential or impulsive, that has strong correlation with a small group of products. When a retailer wants to change a product location, it is important to understand how it will influence other products, and in case it is is a non-essential product, such change will not influence the overall store, but should be changed to a strategic location to increase sales of the its group of correlated products.
Although customer segmentation is not the focus on this post, the customer that purchases the Product 27 and its high correlated products, are part of a specific customer segment. Now that we know the purchase patterns related with Product 27 on the store in UK, which is ideal, let's analyse how the it is behaving on the store in France.
You can see that the Product 27 in France only has a strong correlation with one product (ocean blue) and a small correlation with other one (on the 3rd left shelf, barely visible). For sure there is something wrong with the sales in France for that product. We use two more functions available from AssetFloow's API called CustomerSegment() on the sales data from UK, and then DynamicMultiCorr() on both stores. As the name suggest, the first function gets the customer segments, and the second one compares the purchases behaviours of the same customer segments between these two stores, to find when and which combinations of products change their purchase patterns.
During the whole year, the store in France had 376 transactions, let's see the when the problem occurred:
Product 27 should have be sold 7 more times during the year (1,9%), which represents a lost of €125,58 in sales, for that product in a single store, more specifically during the days 2010/12/01, 2010/12/05, 2010/12/07 and 2010/12/09.
Unfortunately, we do not have the context for the problem only based on this sales data. It could be caused by one (or more) of the following: bad location of the product, bad promotion/marketing (potential customer segment doesn't know about this product), missing inventory.
Feel free to reach us for any question or to try AssetFloow with your data. On the next post, we will explore the paths and shopping times that each customer segment does inside a retail store.
Source of the sales data:
Dr Daqing Chen, Director: Public Analytics group. chend '@' lsbu.ac.uk, School of Engineering, London South Bank University, London SE1 0AA, UK.
Whitepapers about AssetFloow intelligence applied to retail stores and warehouses.
#AI #Retail #DarkStores #SupplyChain #Warehouses #CustomerAnalytics