Insights

Ulabox is a Spanish online grocery supermarket. In 2017, it bagged more than €1 million per month (or annual revenue of more than €12 million) and asserts to have more than +95% Customer Satisfaction. The data provided by them had +30000 orders which represented approximately +10000 customers. 

Ulabox served 8 categories of products in 2017, and seven of the eight categories were served across Spain. The fresh Food category was served ONLY in 2 cities i.e. Madrid & Barcelona. 

In this case, I deep-dived into Ulabox data from the supply chain’s perspective and using machine learning tried to decipher their pricing strategy and consumer behavior. This was an “analytics” exercise in its entirety. The parameters were as follows:

• First, the furnished data was anonymized by the company (Ulabox) and they only talked in numbers to maintain the privacy and propriety of the company.

• Second, the parameters of the data were quite generic in nature and corresponded to the basics of timestamps.

• Third, all essential consumer activities (though limited in the description) were presented in percentages. 

The data limitations made me take some supply chain based assumptions (for an e-commerce marketplace) for the sake of this analysis. Based on the assumptions taken, the in-built constraints of data, and the parameters labeled in the data many things were calculated like the average revenue of the company, the average selling price per item sold, or the customer service level of +95% a viable option or is it constraining the capital invested in the company, etc. 

The next step in the analysis was feature engineering. That said, it was an attempt to use the imagination and give a direction to your analysis. I created features with respect to customers, products, and clusters and then merged them with price per item (assumption-based), customer activity on the marketplace, and customers’ geographic location. 

This exercise helped me discover various patterns in their marketplace activity: the impact of discounts on product categories and their numbers purchased, product categories per geography, consumer buying behavior, and most importantly Ulabox’s probable products discount strategy and product category strategy by geography, etc. 

Lastly, post-publication of the work Ulabox (or Sezamo brand) team gave their inputs and their Lead Senior Developer appreciated the analysis. My gratitude to Ulabox for their input that validated my approach taken and the conclusions made.

Thanks for reading this article. 

Wish you Best