Custom systems require a different strategy and approach. As noted above, they should only be considered after other options are ruled out. Some examples that might qualify for a custom system are: extremely dynamic, large job shop situations; crew-based scheduling where there are many subjective constraints, many crucial interdependencies between jobs and resources; and focused factory control where there is one person who is dispatcher, scheduler, and planner. When custom routes are taken, there should not be just one or two small gaps between problem and solutions, but gaping chasms.

If you are going to go the custom route, the choice of developer will be crucial. Try to find one who understands production control, will work in a rapid prototyping fashion, and understands enough about software to build a flexible and robust system. For a reasonably sized situation, it is possible to craft a prototype in one to two months; this will depend on tools, skill, and starting point. It will then take another two to three months to tailor and fine tune the system. This kind of performance has been achieved for tools at the detailed dispatching and scheduling level. You can also hear of implementations taking years and then being cancelled. The effort required and implementation duration is very variable for any decision support tool in production control. Supply chain systems are not any easier. They may take longer or less time depending on the level of detail and coordination desired.

It will help if the...