COMPANIES ALREADY ARE USING data from social media to fine-tune marketing strategies and reach their customer bases. But soon they might also use that data to make operations decisions about what kinds of items to produce or which ones to discount. When companies feed information about social media interactions into prediction models, they can forecast purchases that will be made the following week.
That’s the premise of research conducted by Ruomeng Cui of Goizueta Business School at Emory University in Atlanta, Georgia; Santiago Gallino of Tuck Business School at Dartmouth College in Hanover, New Hampshire; Antonio Moreno-Garcia of the Kellogg School of Management at Northwestern University in Evanston, Illinois; and Dennis J. Zhang of the Olin Business School at Washington University in St. Louis, Missouri. They worked with an online clothing company that had more than 300,000 followers on its Facebook page when they conducted the study. At the time, the company predicted its sales by using metrics based on weekly and seasonal patterns.
The researchers created software that would extract information about the company’s Facebook posts during a six-month period, categorizing each comment as positive, negative, or neutral. The researchers also consulted internal data on the company’s sales and advertising patterns. Then they created two forecasts, one based only on the company’s internal information, and one that added in social media data. The forecast that combined the two sources of information was most accurate, with a mean absolute percentage error of only 5 percent to 7 percent.
The researchers believe that companies can use such social media information to determine what products are of the most interest to customers—for instance, to see which colors of particular items of clothing generate the most positive response—and then ramp up production of those items. Social media data also could help firms decide what products might be most popular in different geographic areas. Researchers note that companies could deliberately create social media posts that will help them make operational decisions— for example, by displaying two different products and using customer response to determine which one to produce. “The Operational Value of Social Media Information” appeared April 20, 2017, in Production and Operations Management.