A recent study has found that social media data can be used to forecast the sales of clothing and footwear items, based on discussions and engagements centered around color. 

The study also found that shifts in fashion demand are primarily influenced by organic shifts in consumer preferences, as opposed to being predominantly dictated by trends set by the fashion industry.

The study's key finding is that finely-grained social media insights can be used to predict color and fit preferences well ahead of the sales season. This information can be used to guide the decision-making process for initial shipment quantities, which is a critical decision for fashion retailers.

The study also found that incorporating social media data into forecasting models can improve the accuracy of predictions by 24% to 57%. This suggests that social media data can be a valuable tool for fashion retailers looking to improve efficiency and boost revenue.

The study was conducted by researchers at the University of Pennsylvania and involved three multinational retailers. The researchers used a variety of machine learning models to analyze social media data and predict the sales of clothing and footwear items.

The study's findings have important implications for the fashion industry. They suggest that fashion retailers can use social media data to gain a better understanding of consumer preferences and make more informed decisions about product design, marketing, and inventory management. This could lead to increased efficiency and revenue for fashion retailers.