Providing good customer service is crucial to many commercial organizations. There are different means through which the service can be provided, such as Ecommerce, call centres or face-to-face. Although some service is provided through electronic or telephone-based interaction, it is common that the service is provided through human agents. In addition, many customer service interactions also involve a computer, for example, an information system where a travel agent finds suitable flights. This thesis seeks to understand the three channels of customer service interactions between the agent, customer and computer: Customer-Agent-Computer Interaction (CACI). A set of ethnographic studies were conducted at call centres to gain an initial understanding of CACI and to investigate the customer-computer channel. The findings revealed that CACI is more complicated than traditional CHI, because there is a second person, the customer, involved in the interaction. For example, the agent provides a lot of feedback about the computer to the customer, such as, “I am waiting for the computer” Laboratory experiments were conducted to investigate the customer-computer channel by adding non-verbal auditory feedback about the computer directly to the customers. The findings showed only a small insignificant difference in task completion time and subjective satisfaction. There were indications that there was an improvement in flow of communication. Experiments were conducted to investigate how the two humans interact over two different communication modes: face-to-face and telephone. Findings showed that there was a significantly shorter task completion time via telephone. There was also a difference in style of communication, with face-to-face having more single activities, such as, talking only, while in the telephone condition there were more dual activities, for instance talking while also searching. There was only a small difference in subjective satisfaction. To investigate if the findings from the laboratory experiment also held in a real situation and to identify potential improvement areas, a series of studies were conducted: observations and interviews at multiple travel agencies, one focus group and a proof of concept study at one travel agency. The findings confirmed the results from the laboratory experiments. A number of potential interface improvements were also identified, such as, a history mechanism and sharing part of the computer screen with the customer at the agent's discretion. The results from the work in this thesis suggest that telephone interaction, although containing fewer cues, is not necessarily an impoverished mode of communication. Telephone interaction is less time consuming and more task-focused. Further, adding non-verbal auditory feedback did not enhance the interaction. The findings also suggest that customer service CACI is inherently different in nature and that there are additional complications with traditional CHI issues.