Book Contents

Ch. 8
Implementing Communications-Driven and Group Decision Support Systems

Chapter Contents
Previous Page
Next Page

Case Study - IBM and GroupSystems

In early 1987, IBM combined efforts with the University of Arizona to implement a Group Decision Support System called GroupSystems. GroupSystems was the result of a research and prototype development project by the MIS department at the University of Arizona. The project was designed to take advantage of the University of Arizona's development team, while IBM developed the organizational structure and methods to implement GroupSystems in a real business world environment.

GroupSystems utilized a set of flexible software tools within a local area network to facilitate problem-solving techniques including brainstorming, idea organization, alternative generation, and alternative selection. The GroupSystems hardware, software and methodologies are combined in specially developed group facilities called decision support centers (DSC). These rooms were 26 feet by 30 feet and contained 11 PCs connected by a LAN to a large screen projector. The PC workstations were placed in a U-shape around the screen. During a session participants contributed ideas anonymously and all at the same time from their own workstation.

A typical session began with the planning stage where the initiator identifies who the proposed group members will be. The group should consist of a variety of people from various business levels and business divisions. The initiator will also identify what the fundamental focus and purpose is of the session at hand, and any limits or constraints that need to be placed on ideas such as, no increase in employees. The final step in planning is to send out invitations with the agenda and logistical information detailed.

At the beginning of the actual meeting the initiator reads the agenda statement aloud. Typically the activities begin with a brainstorming question where members freely contribute their anonymous responses on the computer. Then the participants identify a list factors that should be used to judge the ideas. The members will prioritize these factors by a ballot process. After the conclusion of the session the initiator receives a hard copy of all the comments generated and uses this information to make a decision.

GroupSystems at IBM had four types of participant relationship sessions. The first is intra-organizational, which is composed of IBM employees from the same functional area. The second type is inter-organizational, within the company, where horizontal groups of employees from different functional and executive areas are group members. Then going outside the company is the inter-enterprise group between IBM and its business associates including, suppliers and vendors. The last group is inter-enterprise between IBM and its customers.

IBM and the University of Arizona reported success with the use of the GroupSystems application. The most prominent benefit reported was saving time (an estimated 56% in work/meeting hours). IBM estimated that the savings were so great that the return on investment for the facility was one year. The process is highly effective in keeping the agenda on track and reducing the amount of unrelated discussions about outside issues. Plus, with greater importance placed on the planning stage, meetings run much smoother and quicker because the agenda has been predetermined. All of these characteristics result in time savings.

IBM was also been able to resolve issues in a shorter time span, because meetings were longer in length but fewer in number allowing problems to be resolved in a month and a half, on average. Anonymity of participants resulted in many positives for the company. First, there was greater participation, because respondents did not feel that their careers were in jeopardy because of meeting comments. The increased participation rate carried over into the work culture with members approaching their leaders more often to keep themselves involved with related issues. Also, there was less "group think", because people had less direct influence on others. The system was designed with easy to use graphical interfaces, thus low levels of computer literacy did not deter use of the system. The most surprising benefit of the new GroupSystems was the uses of the session data after a meeting. This data was used to support managerial decisions, to document future sessions, in presentations and project management workshops, for technical reviews, and in bimonthly meetings to keep the group focused on the project until completion.

IBM's development of these decision support environments was intended to be a slow process enabling the company to maintain adequate support. The first Decision Support Center (DSC) was opened in October 1987 at the Oswego, New York site. Managers conducted 26 test sessions during the fourth quarter of 1987, the results showed promise for success. In the second and third quarters of 1988, three additional DSCs were added in at offices in Maryland, Virginia, and North Carolina. In the next 12 months, another 13 sites were added throughout the U.S. and Canada. By mid-1990, IBM had 21 Decision Support Center sites with GroupSystems capabilities.

The above summary was prepared by Nikole S. Hackett and D. J. Power and is based on an article by McGoff, C., Hunt, A., Vogel D., and Nunamaker, J. titled "IBM's Experiences with GroupSystems" in Interfaces, November-December 1990, pp. 39-52 and on discussions with some of the researchers and staff at IBM and the University of Arizona.

Questions for discussion:

  1. What benefits resulted from using GroupSystems at IBM?
  2. What types of participant relationship sessions were held using GroupSystems?
  3. What is your reaction to the typical GDSS session held in an IBM Decision Support Center?
  4. Does IBM still conduct GDSS sessions?


DSSResources.COMsm is maintained and all its pages are copyrighted (c) 1995-2002 by D. J. Power (see home page). Please contact This page was last modified Wednesday, May 30, 2007. See disclaimer and privacy statement.