in our factory so we can devote our energy to making the new system successful. Collins' Views Baxter also talked with Collins, who argued that BMC should continue its process of building the manufacturing systems that it needed. He estimated that the needed systems could be completed in about two years at a cost of around $420,000$220,000 for outside help (including training his people in new development tools) and $200,000 in internal costs. When Baxter asked Collins why BMC should not purchase the EMS software, Collins replied: First, the EMS software is far more complicated than we need. For most general manufacturers each part may require six operations on six different types of machines, and each part has a routing that is different than other parts. Then several parts may be assembled into a subassembly, so you have two- or three-level bills of material. We typically take a coil of steel, stamp out the part, clean it, box it, and ship it out, so both our routings and our bills of material are very simple, as is our production process. The EMS system is designed for much more complex manufacturing. Second, we have had little or no experience with computerized production systems. Does it make sense for us to try to jump to a very complex and sophisticated system like the EMS proposal? Lucas has a very superficial understanding of this software package, and he doesn't know any of the details of how it will work. Therefore, he has no idea of the difficulties that his people will run into in adapting to this complex package. It will require them to do many tasks that they have never done, or even considered doing. And they don't need this complexity. Wouldn't it be better to build our own systems that correspond to where we are on the learning curve and plan to upgrade them as we progress in our understanding of our systems needs? Third, it is likely that the system does not fit the way we are running the business. Do we change the system or do we change how we run our busi- He may or may not make the changes that you want, and in fact he may make some changes that you do not want. If you do not expect the system to change and it is a common system, you probably should purchase it. For example, one general ledger system is just like any other, and they haven't changed in 20 years, so you should purchase this application. But we are continually changing things out in the shop, and if we build our own systems, we can change them when we need to. Finally, we have demonstrated that we can build and successfully install our own systems, but our record with purchasing and installing manufacturing systems is dismal. The EMS proposal may fulfill our needs, but then again it may not. We failed twice in the past because the system we purchased did not fit our needs. Why take that chance again? "You seem very concerned that the EMS system might not suit our needs or that our needs might change," Baxter replied. "Could we modify this system if it does not suit our needs?"' Collins said: Because we will not have a source-code version of the software, it will not be feasible for our programmers to modify the functionality of this system. However, we can write interface software to change the form of the system's input and output. When Baxter noted the cost and time differences between purchasing and building the system as estimated by Moore, Collins replied: The figures Lucas quotes are very misleading. The purchase price is but a part of the total cost of buying, installing, and maintaining the software. To be sure you are choosing software that truly meets your needs, you must put a substantial effort into defining your needs and evaluating each candidate package against those needs. One of the major weaknesses of the present proposal is that this process of defining needs and evaluating possible packages has been completely ignored. In my opinion we must go through this process before buying any packaged software, and this will affect both the proposed cost to software, and it was a terrible experience. Each vendor claimed that his software would do anything you wanted to do, and there were so many questions we should have asked but didn't. Vendors all offered integrated packages that included production scheduling, but you also got sales, inventory, purchasing, shipping, etc. We made our selection and paid about $120,000 for the system, including both hardware and software, which was a large expenditure for us at the time. Then we started to load the data and implement the scheduling package. The training the vendor provided was poor, the manual was full of errors, and support from the vendor was minimal. We worked and worked, and finally became so frustrated by our inability to get the system to do what we wanted it to that we just gave up. On top of everything else the vendor went bankrupt. It was a total disaster- $120,000 down the tube! As mentioned previously, by 1991 the problems in meeting shipping schedules had gotten so bad that BMC began to have to turn down new business. Management again decided that they had to do something about machine scheduling, so again they decided to purchase a scheduling package. Sue Barkley remembers: This time things went better. Nancy Shaw and I got more people involved in the decision on what package to buy. This vendor provided some in-depth training to our MIS people, and vendor people came down here for 2 weeks to help us load the data and get the production scheduling module working. Again, we found that the manual was full of errors and that the vendor people did not fully understand the logic that the system was using. But we got the system up and working and taught the production scheduling people how to use it. The problem was that whenever we had to expedite something-give it top priority because it had to be shipped quickly - the schedule had to be regenerated, and that took 2 hours. Then we had to take the schedule for each machine and examine it to see what the impact on its schedule was and change what it was going to do. Because we were always expediting something, we were constantly chuming. After about a month the production scheduler came to me and said, " Tm not getting anything done. It takes me 2 hours to regenerate a schedule. I look at it and I then have to change five or six machines because of what the system did. Then it takes me 2 more hours to generate a new schedule and I have to change another five machines, and I have to go through the cycle again. It's just a continuous process of change, change, change!" We tried for another month to make the system work for us, but we were in such bad shape with our capacity that we just couldn't take the time to try to cope with the system anymore. So we abandoned it and went back to our Lotus spreadsheet. The $150,000 that we had spent for that system was down the drain! The Present MIS Department In 1994 Shaw left and BMC hired Don Collins to replace her as MIS manager. Collins had 20 years of experience as a lead systems analyst with a large manufacturer and broad experience with manufacturing systems. In 1996, Collins has a programming staff of four. The 1996 capital budget for hardware, software, and other information technology items was about $200,000. The MIS expense budget for payroll, supplies, and education was about $350,000. The MIS department is using a development tool called Cyber Query Cyber Screen (CQCS) from Cyber Science, but Collins is giving some thought to what BMC's development environment of the future should be. The Data General MV computer is becoming obsolete and is reaching capacity, so BMC will have to obtain additional capacity soon. In order to plan a production schedule you need to know what you have in inventory, so the MIS group has created systems to track raw-material, in-process, and finished-goods inventories. MIS has also developed a minicomputer system that accepts EDI orders from customers and allows the customer service group to create a shipping schedule on the computer. Collins believes that within 2 more years the MIS group can build and install a set of manufacturing systems that will satisfy BMC's basic needs and provide quite satisfactory EDI service to customers. This success in building new systems opened BMC managers' eyes to the possibilities for using the computer, and they have generated so many requests for new systems that an MIS steering committee has been established to approve projects and set systems development priorities. The members of the MIS steering committee are President Kyle Baxter, Controller Lou Wilcox, Sue Barkley, and Don Collins. The New Proposal In late 1996 Lucas Moore, vice president of manufacturing, suggested that BMC purchase and install an integrated package of manufacturing software sold by Effective Management Systems, Inc. (EMS). Moore had worked as an engineer with the company for 7 years and then took a leave for 2 years to get an MBA. The vice president of manufacturing retired soon after Moore returned, and Moore was promoted to that management position. Moore supports the proposal that BMC install the EMS Time Critical Manufacturing package consisting of eight modules: shop floor control, EDI integration, inventory management, factory data collection, standard routings, labor collection, engineered product configurator, and general ledger. The purchase price of this software package is $220,000, including documentation, training by EMS, and consulting help during installation of the software. The cost of a software maintenance contract is $55,000 a year, and EMS will make limited changes requested by BMC at a cost of $60 per hour. The EMS software will run on several minicomputers, including BMC's Data General MV. However, additional computer capacity will be needed whether BMC purchases the EMS package or builds its own manufacturing systems. Moore's Views Moore is relatively new to the manufacturing area, having taken over that area about a year ago, and was not involved in the past attempts to purchase scheduling software. Moore explained to Baxter that BMC should purchase the EMS package for the following reasons: We are still fudging our EDI and still scheduling with a Lotus spreadsheet. The entire industry has passed us by in our use of the computer in manufacturing and we are in danger of losing our reputation as a world-class parts manufacturer. Both my MBA studies and our experience with the new inventory systems that Don has installed have convinced me that computer systems can significantly enhance our efficiency and improve our service to our customers, but we can't wait another 2 years to complete home-grown manufacturing systems that will still need to be upgraded before they are really first class. I have had extensive discussions with EMS manufacturing specialists, read their literature, and seen the proposed systems demonstrated, and am convinced that the proposed system will do everything that we will ever want to do. EMS has assured me that there will be no problem integrating these manufacturing modules with our existing financial systems, and that we can be up and running with the entire system in 6 months. "Given that our MIS group is doing a good job developing new systems," Baxter asked, "why should we purchase the EMS package rather than build manufacturing systems in-house?" Moore's reply was: The time and cost differences between purchasing and building are too significant to ignore: 6 months to install this advanced system versus 2 years to build our own basic system, and a firm $220,000 to purchase this system versus over $400,000 to build our own. These costs do not include new hardware, but we will need to increase our capacity whether we purchase or build our new systems. Furthermore, we will get a high-quality state-of-the-art system instead of a simple "first try" system. EMS has sold this system to hundreds of manufacturers, and thus has been able to spend much more time and money developing it than we could possibly afford. EMS has a large staff of more creative and sophisticated programmers than we can get, and EMS has gone through several cycles of improvement of this system based upon the experience of hundreds of users of the earlier versions of the system. It is true that the EMS system will not always do things the way we currently do them. But is the way we do them better than the way that is based on the experience of hundreds of manufacturers? We are always making changes in how we do things, so it will not be difficult for us to make some changes to conform to this new software, and I expect that these changes will improve our operations. "We have not been successful in two tries to use purchased software packages in the manufacturing area," Baxter noted. "What makes you think that we would be successful this time?" Moore replied: There are a number of important differences this time. First, in the past there was little ownership of the new system by the factory people, but this time I am the champion of the new system and my people will make it work. Second, in the past the conversion strategy was flawed-BMC tried to install scheduling without having inventory data under control, but this time we will go at it a module at a time in the sequence that EMS has been very successful with in many previous installations. Third, during the previous attempts we were pushing capacity and no scheduling system was going to work when we were having to expedite everything, but today capacity is not a major problem and things are reasonably calm additional systems to translate from one packaged system to the other packaged system. In addition there will be costs of training the users, data conversion, and the changeover to the new system. A good rule of thumb for the total cost of installing a purchased package would be twice the purchase price of the software, which in this case would be $440,000. I doubt that we could do it for any less, and that compares with about $420,000 to build our own systems, which includes all the costs involved, such as training, conversion, and defining the needs of our manufacturing people. It will take at least a year to properly evaluate and install a purchased system. This is less than the 2 years we will need to complete our own system, but we will be installing and using components of the new system as we complete them, so the time advantage is not that great. When asked what it would take to do a more complete evaluation of the proposal to purchase the EMS system, Collins replied: We would need to spend about 6 months studying our manufacturing area to determine what we are doing now and what the new systems should do. Then we would take some time to explore the many packages that are available, and winnow them down to the three or four most suitable. Then we would invite the chosen vendors to submit proposals so we could study and evaluate each of these proposals in detail and pick the best one. Meantime, we would prepare a proposal for building the new system that would describe the proposed system in detail and include a plan for its development including schedules of both time and dollars. Finally, we would compare the best proposal with the plan for building the system ourselves and decide which to do. That would take at least a year and cost between $50,000 and $90,000. Decision Time After his discussions with Moore and Collins, Baxter sat down with his sister, Sue Barkley, to discuss what to do about Moore's proposal. "Sue," Baxter began, "you were able to get the second manufacturing software system we bought up and running, but conditions in the shop were so chaotic that we abandoned trying to use it. Why don't we go back and try it again?" Sue replied: We recently considered trying again to use this system, but the special computer we bought to run it died and the software vendor has gone out of business, so we were out of luck. "Lucas claims that BMC is losing its reputation as a world-class parts manufacturer because its systems are inadequate, and therefore BMC must purchase a system without delay," Baxter said. "Do you believe that it is critical that we get these new systems immediately?" Sue thought a while before replying: I don't think that our customers care about our systems as long as we provide high-quality products at a good price and deliver them when they are needed, which we are doing. From their perspective, we are already interacting with them via EDI, so that is a problem for us rather than for them. It would be great to have the proposed systems as soon as possible, but we have been getting along without them for a long time. "Well, Sue," Baxter said, "I still don't know what we should do. What do you think