Social interactions have been established as a means to help promotions and sales for manufacturers and retailers. Word-of-mouth (WOM), in particular, is proven to increase awareness and drive purchases. Given that small offline retailers have limited resources in marketing, online WOM can play a key role for their offline business performance. In this paper, we focus on two types of online WOM, public discussion and social referral, and study their generation processes by taking into account the multichannel context of both online purchases and offline consumption. To this end, we combine data from three sources: product (or deal)-level sales from a major deal site in South Korea, social interaction records collected by web crawling, and retail revenues at a district level from Korea National Statistical Office. We use a multivariate poison lognormal model to estimate three equations in the same structure with correlated errors, which only differs by the following dependent variables: number of product discussion, number of social referral, and the number of social coupons sold. Our empirical analyses suggest the following. First, the two types of WOM respond in opposite directions for the influencers in the multichannel sales context: the greater number of co-located online deals decreases public discussion but increases social referral. Next, the larger offline retail size increases public discussion, but has no significant effect on social referral. Finally, the results provide practical insights that small offline retailers can improve sales in the multichannel context by effectively managing the generation of different types of online WOM.

• Jikyung (Jeanne) Kim(IE University, Spain)
• Sanghwa Kim(Yonsei University, Republic of Korea)
• Jeonghye Choi(Yonsei University, Republic of Korea)