Rob Karel and Keith Gile of Forrester recently published a report on "Turning Transactions Into Decisions". This was mostly about data integration options in operational BI situations but as I read it I found some interesting tidbits. They make the point that the information being integrated/delivered must facilitate decision-making and that there are typically many decision points in a process. I could not agree more - indeed this focus on decision services within a process is key to enterprise decision management. As they continue on they draw a distinction between manual decisions (that require BI support) and automated ones. However their example of manual decision - deciding about credit-line increases - is actually a prime example of a decision that is increasingly automated in an EDM-like way! Indeed I have blogged about using rules and analytics to automate credit decisions at a UK bank and about how similar loan decisions can also be automated. Their other example, routing of calls, is a typical automated decision and one that can also be improved using EDM.
Now when you are talking about automated decisions, you need executable rules (procedures, policies, regulations) and analytic insight (embedded into decision execution) to replace the policy manuals and reports that would support a manual decision. This is why EDM uses both rules and analytics to focus on automating and improving operational decisions. If the frequency of your decision is high and the time for taking the decision is short then automate it using EDM.
Another point to note here is that using analytic techniques to build the rules you need can mean not needing so much data at execution time - you process the high data volumes/multiple data sources to build the models but the models may not require so much data. For example, there was a great story in Blink about the small number of data elements that actually made a difference to a specific medical decision. Using an EDM approach might actually reduce the need for data integration in your operational systems. The article also makes some good points about data latency (see my comment on a Richard Hackathorn article) and made me think of this post on leveraging customer intelligence through decisioning.