An empirical analysis of software life spans to determine the planning horizon for new software

This paper presents an empirical analysis of the life span of over 180 systems aimed at developing a model for determining the planning horizon for new software at the business case stage of software acquisition. At this early stage, the firm has limited knowledge about the project, but must make crucial decisions, such as scope (breadth of requirements), approach (both insource vs. outsource and custom vs. package) and technology, including fit with standards (adhere to current vs. adopt new technology). These decisions are associated with different system lifetimes that, in turn, impact both the costs incurred and benefits received from the system. The failure to explicitly and properly address these differences can lead to the implementation of systems better left undone or to unintended consequences, such as the Y2K problem. We find that technology and approach, but not scope decisions are strongly related to system lifetime. In particular, systems that use an operating system that conforms to the firm’s standard or are acquired using a blended team entail longer system life. On the other hand, shorter system life is indicated if the system is technically complex, custom developed or uses an older programming language. Furthermore, modified packaged software is shorter lived than is a vanilla package. In addition, environmental variables also impact the appropriate horizon. For example, as one would expect, strategic systems are used longer. On the other hand, somewhat surprisingly, systems sponsored by executives last less long and despite the quickening pace of technological and business process advancement, a small trend toward longer lived systems is uncovered.