This paper examines an important issue emerging in information systems management--the decision to proceed with an expert system application in a business setting. The focus is knowledge based systems at the lower end of the complexity spectrum--small, very focused systems that can be implemented by the use of shell-based development environments. This group represents the majority of expert systems that are currently being implemented and has some characteristics quite different from the larger systems. A classification scheme is suggested to differentiate three levels of ES development, from multi-million dollar life cycle cost ES environments to those that are in the low five figure range. The Low End segment of the range, the focus of this paper, is characterized by lower unit costs, powerful development tools and a large number of small, successful applications. The important role of Low End systems is discussed, with particular emphasis on their relatively high yield in standalone applications. Such systems do not meet the AI demands of moderately or very complex problems but there is a surprising breadth in their use. A group of key success factors for Low End systems is proposed, based on a synthesis of the applications literature. To operationalize these factors, three actual cases using Low End technology--from marketing, government and agribusiness-- are briefly described. Low End systems are not all gain. Their low unit costs can often mask the risks of proceeding headlong into an application without careful examination of the variables that can predict successful results. An agenda for action is offered for specific management policies for the planning of knowledge-based applications.