AUTOMOBILE SALES FORECASTING USING TWITTER

Social listening on Twitter is actively performed as alternative data for marketing and product development in various business fields. In an attempt to create a formula to forecast future automobile sales from content posted on Twitter, we collected and analyzed tweets related to automobiles in Japan from June 2018 to May 2020. The results showed that automobile sales could be forecast when the number of tweets was high. This indicated that social media is useful as alternative data for product sales forecasting. During the study, we noticed that forecasts became possible after a product entered the so-called "cash cow" phase in the "product portfolio matrix", developed by the Boston Consulting Group, which uses market growth rates and relative markets to develop business strategies. Based on this, we propose a hypothesis that in order for the sale of a product to be forecast, the product needs to enter the “cash cow” phase in addition to a large number of tweets.