[1F5-OS-4] OS-4

[1F5-OS-4-01] Time series analysis using PR measurement framework

〇Kunihiko Ueshima¹, Taito Tosaka², Kohei Taniguchi³, Teruaki Hayashi⁴, Yukio Ohsawa⁴ (1. Japan Data Exchange, Inc, 2. OZMA, Inc, 3. kikkoman corporation, 4. School of Engineering, The University of Tokyo)

Keywords:Data Flow, PR Measurement, Time-series Data

The democratization of statistics and big data mining has made it easier for many companies to use digital marketing methods. However, these methods tend to focus on understanding the short-term return on investment for a single campaign, it is difficult to understand the long-term PR effects.

In this study, we tried to evaluate the impact of accidental consumer trends and multiple campaigns by companies. For several brands, we create datasets that combines multiple time series data, and perform data observation and time series analysis.

As a result, three points were suggested. First, though the process of developing accidental consumer trends can be investigated in a common framework, the duration and extent of the impact will depend on subsequent communication. Also, a single campaign may not have a direct impact on sales. It is better to have continuous communication across multiple media. Rather than analyzing a single campaign in too much detail, important index should be organized and their linkages evaluated.

If we apply this result to other brands, more advanced PR strategies will be created.