(Posted by guest blogger, James Taylor)
- Using the reporting interface
One of the critical aspects of Enterprise Decision Management is not to focus on reporting as the reason for having and analyzing data. This is not to say it is not important to report on data, it can be, but to remember that reporting is not the be-all and end-all. Predictive analytics does not mean predictive reporting.
- Focus on insight not on action
Even my friends at Juice are focused on delivering insight to people. EDM, in contrast, is focused on delivering actions. While you often want to use some of the same techniques to deliver supporting insight, you want the "last mile" focused on the actions that the system is trying to take or cause to be taken.
- Interfaces for normality not exceptions
Normal transactions, the day to day, should be fully automated and only exceptions should be presented to people for review. This means designing the last mile to make it easy for people to know when there is an exception for them to handle and then using the kind of techniques the folks at Juice discuss on their blog to make sure those people can act effectively for these exceptional transactions. Don't design for the transactions your decision service can handle, design it for the others.
- Too much context
Not too much context for a person handling an exception but too much context when automating a decision. After all you want to be able to use the decision in multiple processes, multiple systems. Making the decision service stateful (not stateless) or tying it to the context in which you first use it will severely limit its reuse.
Remember, you are trying to automate decisions not just support them.