Client
A US-based fashion retailer with physical and online stores across the United States and abroad.
OPPORTUNITY
The client was planning the launch of a new collection of fashionable sportswear across its stores in the US, catering to its main customer base - upper middle-class Generation X consumers. Although having an onshore analytics team available, the client tasked SG Analytics with the creation of an independent sales forecast model to predict the turnover at the store and regional levels.
SOLUTION
SG Analytics’ data scientists and retail experts decided to build a mixed model that combined autoregressive and economic features to account for economic factors as well as for proxies for brand perception and consumer trends.
SG Analytics’ initial model included 3 broad categories of factors:
-
Microeconomic and product-related factors, including product category, launch season, price, and marketing spend.
-
Macroeconomic factors including average personal wealth, average personal income, GDP growth, CPI (Consumer price index), CCI (Consumer confidence index), etc.
-
Previous sales across relevant product categories.
SG Analytics conducted interviews with select experts from research and industry to cross verify the set of factors and get reference points for their parameters.
The client provided SG Analytics access to their marketing, sales and customer records. SG Analytics gathered the remaining macro and microeconomic data from free and paid data sources. SG Analytics cleaned and standardized all collected data for further processing.
Based on the data SG Analytics created a predictive model that would forecast the sales of new collections under consideration of various microeconomic, macroeconomic, and autoregressive factors.
Value Delivered
1
SG Analytics delivered a predictive model to forecast the turnover of the clients’ upcoming collection of fashionable sportswear. Due to the model’s flexibility, it could be deployed for other future collection launches as well.
2
The model was largely independent of the clients in-house model and delivered more accurate forecasts, especially on the store level.
