Businesses as Buildings: Metrics for the Architectural Quality of Internet Businesses

Abstract

Metrics for the architectural quality of Internet businesses are essential in gauging the success and failure of e-commerce. This study proposes six dimensions of architectural metrics for Internet businesses: internal stability, external security, information gathering, order processing, system interface, and communication interface. The metrics are based on the three constructs that have been used to evaluate buildings in the real world. The structural construct indicates that Internet businesses need to be stable internally and secure externally. The functional construct implies that Internet businesses should provide convenient functions in the information-gathering and order-processing phases. Finally, the representational construct indicates that they need to provide a pleasant interface both to the systemand to those using it. For each of the six metrics, we have constructed questionnaires to measure the perceived level of architectural quality and identified feature lists that may be closely related to the perceived quality level. Large-scale empirical studies were conducted both to validate the proposed metrics and to explore their relevance across four Internet business domains. The validity of the metrics has been obtained in three ways. First, the content validity of the metrics was assured by pretests and pilot survey. Second, the results from the confirmatory factor analysis showed that the metrics had high convergent and discriminant validities. Finally, the reliability coefficients were found to be high enough to establish the reliability of the proposed metrics. The relevance of the metrics has been explored in two ways. Structural equation models were used to test the causal relations between the three constructs and user satisfaction, as well as customer loyalty, in four domains. Correlation analyses were used to explore the relations between the perceived architectural quality and objective design features in four domains. This paper ends with the implications and limitations of the study results.