This paper describes the development of a guide robot for a shopping center, which attempts to build rapport with customers while providing route guidance and other shopping information. A field trial was conducted at a shopping mall for 25 days to study the possible role of a robot in a shopping center. Since a shopping center is a place where local people repeatedly visit, it is important for the robot to interact with people multiple times. Using passive RFID tags to identify customers, and using the WOZ method to overcome difficulties in speech recognition, the robot built rapport with customers through a series of interactions based on a set of four design principles. During each day of the field trial, the robot interacted with approximately 200 customers, and the robot was able to sustain customers' interest throughout multiple interactions. The robot also provided shopping information through natural conversation with the customers, and the experimental results revealed that 63 out of 235 people in fact went shopping based on advice from the robot. These results seem to demonstrate a promising potential for guide robots designed to build rapport with customers in a shopping mall.