A marketing analysis using massive tiny sensor nodes
Masayuki Iwai, Masato Mori, Hideyuki Touda · 2009
Existing huge e-commerce Web sites such as Amazon and eBay tends to have a recommendation function to stimulate customer's potential needs by analyzing their click information. On the other hand, POS (point of sale) data is analyzed by real life retail store owners to gain information about the purchased items. However, POS data is only useful to analyze information regarding ¿purchased¿ items; i.e., POS-based analysis is not capable of accessing the customer's potential needs. We have deployed a ¿SmartStore¿ system for the retail stores. SmartStore targets hand held sized selling items, such as cellphones and toys. SmartStore gathers customer's action information, such as the act of holding or touching a product attached with massive tiny sensor nodes in the shops. The action data is then interpreted and analyzed as the user interest corresponding to the target product. To show the system scalability, we have gathered more than 3,950,000 photonic and movement sensor data from approximately 120 customers for the duration of 3 days. This experience gained from our experiment leads to the feedbacks obtained from the retail store owners and customers. SmartStore is the first step to converge pervasive computing technology with the ¿real life¿ marketing domain.