Sales volume decomposition – analyze the factors influencing sales volume at a global coffeehouse chain

Our client, one of the world’s largest coffeehouse chains wanted to identify factors contributing to the increase or drop in sales for various products spanning from drinks, food to merchandise across time. 

Thus, understand the precise formula which drives the sales for different products at a Global Coffeehouse Chain .

TransOrg developed a sales volume decomposition model to analyze the factors influencing sales volume  and decompose them to gain insights into the key drivers of sales performance

• Scoping 🡪 Actual sales quantity was modelled at product and store level on daily basis.  Separate models were built for different products, store types and customer base.

• Data Collection 🡪 Study and gather relevant data sets encompassing a wide range of variables, such as promotional activities, product mix, customer loyalty, promotional activities, weather conditions, and seasonal trends.

• Feature Selection 🡪 Data points such as seasonality, big deals, discount percentage, holidays were found to be influencing sales volume for different products.

• Feature Engineering 🡪 Further, new features like free quantity, new registrations, unexplained store promotion and store drop, store specific variables were engineered which were found to be affecting the sale volume.

• Statistical Modelling 🡪 Linear regression was used to model the relationship between selected input features and sales quantity for different products and thus the effect of selected input features were gauged on the sales volume for a particular product.

Key Impacts

• The inherit seasonality in the sales of every product was identified.
• Client was able to gauge the impact of holidays on the sales across products which can be used to introduce different offerings on different holidays.
• Discount sensitivity of every product was identified which helped them to mange their discount campaigns more effectively.

Key Insight

• Key sales metrics, such as total sales volume, sales by product category, sales by store location, and sales trends over time were presented via an interactive dashboard to monitor and understand the sales performance at a glance.