Reproducing stylized facts of word-of-mouth with a naturalistic multi-agent model

Word-of-mouth is the interpersonal process by which information about an innovation diffuses in a social system. Little empirical data is available from the field. Nevertheless, several stylized facts are identified and widely accepted. For instance, consumers actively search for information. It is also known that knowledge about incremental innovations diffuses quicker, because part of the knowledge is already available from previous innovations. Previous models applied to wordof-mouth do not reproduce these stylized facts. We propose a model compliant with these stylized facts, in which knowledge representation relies on associative networks, and whose communication protocol is based on social objects. Simulations illustrate the increased representativeness of the model by modeling active search for information and diffusion of incremental innovations. We also show an important change in the diffusion rate due to active search for information. 1 Modeling word-of-mouth 1.1 Evidence on word-of-mouth In this paper we study word-of-mouth (WOM), that refers to the process by which knowledge about innovations (new products, practices or ideas) diffuses in a social system. Nowadays WOM receives increasing attention, not only to enhance viral marketing efficiency, but also to study the diffusion of innovations, the dynamics of opinions, and any other social processes concerned with interpersonal communication. Data collection for word-of-mouth is hard to set up on a large scale. This explains why so few empirical studies are available (e.g. [1]). Nevertheless, several stylized facts are widely accepted in the field of diffusion of innovations [2], persuasive communication [3], consumer behavior [4] and other domains. Given the lack of quantitative data from the field, these stylized facts are the only basis available to evaluate the plausibility of WOM models. We summarize here several stylized facts (SF) that are pervasive in literature. SF 1. Active search for information. It was shown in various fields (impact of advertisement [5], diffusion of innovations [2] and consumer research [4]) that consumer knowledge about a given innovation evolves in several steps. First the consumer is unaware of the product. Then, he/she becomes aware of the innovation due to some cues received through advertisement or interpersonal communication. Based on these cues and his/her own preferences, he/she can search actively for information by retrieving information from his/her friends. Only then can he/she enter in a decision step. When a consumer has enough information, he/she can also proactively transmit his/her information to his/her acquaintances. This communicative behavior depends on individual characteristics: a given consumer looks for information, speaks proactively and adopts a product depending to its needs, motivations and interests. SF 2. Partial knowledge transmission about topics When two people meet, they discuss various topics that one or the other have in mind. Obviously they cannot exchange all their knowledge, but rather exchange only their beliefs about the topics of the discussion. Moreover, people often meet and discuss various topics but not innovations. During a field study, Carl [1] measured that only 16% of interpersonal communication was about products. SF 3. Previous knowledge New information received by an individual interacts with previously held knowledge. A piece of information can be misunderstood, so consumers don’t fully evaluate the benefits of the innovation [2]. Moreover, a lot of innovations are in fact incremental innovations, which share information with previous products. As pointed out by Rogers [2], incremental innovations diffuse more quickly because part of the knowledge necessary for their adoption is already known from previous innovations.