Most companies view social media platforms such as Facebook through the prism of marketing, however research out of the US points to their use in developing sales forecasts.
Research from the Kellogg School of Management found data derived from social media platforms can help to improve forecasts via the use of advanced algorithms. Researchers were able to incorporate information about an online clothing company’s Facebook interactions into prediction models.
By employing different models, the researchers were able to more accurately estimate purchases the following week.
Software was written to extract information about the Facebook posts of the company, which had more than 300,000 followers at the time of the study. Researchers then used language-processing software to categorise each comment as positive, negative or neutral.
Having also obtained information about the company’s sales and advertising campaigns, the researchers produced a baseline sales forecast, including only internal company information, and a forecast combining internal and social media data.
A variety of prediction methods were employed, mostly relying on machine learning, with accuracy assessed via a measure called mean absolute percentage error (MAPE). It showed how much the estimate deviates from actual sales.
Forecasts including social media data brought MAPE down. However, poorly performing models could deliver accuracy even worse than the baseline model, pointing to the importance of both data and methods.
While the study did not reveal why the Facebook information improves forecasts, Antonio Moreno, Kellogg associate professor of operations, speculated it may be a reflection of how much attention customers are paying to the brand, along with good or bad word-of-mouth.
“By introducing social media data, we can do better,” Moreno commented.
“But it looks like the first step should be having better methods.”
The study points to the potential for companies to be more strategic with social media posts, potentially employing posts to elicit specific information to help guide operations, such as displaying potential products and then making decisions based on social media users’ reactions.
|Passionate about the state of Australian small business? Join the Smarts Collective and be a part of the conversation.|