Infrastructure maintenance is a complex task due to operational needs of service quality, availability demands, traffic intensity, and budget constraints. Traditionally, maintenance decisions are based on infrastructure managers' experiences, judgments, and available choices. Though, the technology push and the availability of an abundance of data have made an urge to derive the decisions and insights from data. This data-driven approach resulted in computerized solutions, e.g. decision support system (DSS), which are rich in data but poor in insights. The DSS confronted with data management challenges of data acquisitions, data cleaning, and data reliability further complicate the already complex task of maintenance decision-making. To tackle these challenges of maintenance decision-making and data management, a decision-driven approach is suggested for the development of DSS. The decision-driven approach builds on the definition of decision context specified by infrastructure manager. The data requirements are provided by decision context, where the interrelationships between the data and the decision context are made explicit by developing an information model. We illustrate the decision-driven approach for DSS development using a case study of maintenance decisions for bridge selection. It is found that the decision-driven approach directs the focus towards the decision context definition and decision analysis while minimizing the overhead of data management.