5th Generation Bam?
February 28th, 2007
Sandy Kemsley’s blog on business process management on eBizQ reports on a Presentation given by Gartner Analyst Bill Gassman hilighting the difference between BI and BAM. She comments that
“So what’s the difference between BI and BAM? According to Gassman, BI is used for insight and planning, and is based on historical — rather than real-time — data. BAM is event driven, and issues alerts when events occur. Personally, I think that there’s a spectrum between his definitions of BI and BAM, and it’s not clear to me that it’s a useful distinction; in many cases, data is trickle-fed from operational systems to BI systems so that the data is near-real-time, allowing dashboards to be driven directly from the BI system. True, traditional BI tools will typically see update intervals more like 15 minutes than the near-real-time event alerts that you’ll find in BAM, but that’s not a problem in some cases.”
The problem with definitions such as “BI” and “BAM” is that real life is inevitably more grey and less black and white. Bill does a good job at classifying vendors to provide separation and distinction between vendors, but perhaps I can add a twist to the definitions which may help.
In the context of BPM, it’s useful to think about an additional dimension which has not been discussed here: when do you need to do the analysis: design time vs run time?
Rather than trying to look at what is real time and what is ‘near real time’ which is think is less useful, I always suggest that a good starting point in optimizing processes is to consider when the optimization needs to be done. Is the requirement to do it while the process is running on a continuous basis, or are you seeking to understand how well your processes executed so that future instances can be tuned?
Querying process logs to produce reports and dashboards can help business analysts to tune processes, but is an “after the fact” (design time) activity. This is traditional BI applied in a BPM context and there are several notable examples of BI vendors collaborating with BPM vendors to facilitate this.
If you ask users what they think BAM is, then they will describe a process oriented dashboard that can give visibility into processes executing in real time. There is real value for process analysts to get visibility of bottlenecks of processes modelled in BPM tools.
When you want to optimize processes on a continuous basis, based upon the process instance in flight, then this needs to be done automatically, and in real time. For example, you might want to reconcile both supply and demand in real time in order to adjust service levels or adjust prices dynamically.
A real time dashboard alone won’t achieve this goal, and part (or all) of the processes and the data needed to do this are often not modelled in a BPM tool. So these types of systems tend to be data centric (not necessarily BPM centric), and event driven.
The characteristics also are of more sophisticated analytics which are automating an analysis process to detect exceptions, calculate risks and forecast, say, delivery times. In order to do this often a significant quantity of both real time and historical data is required, but processing this in real time is beyond the capability of a BI tool or even database. The results of these analytics are used to then initiate processes, or provide information to modelled processes as an analytic process step. Oh, and of course this all has to happen in real time.
Building ‘business intelligence’ (in its broadest definition) into processes is clearly an obvious way of making your processes smarter. Is it 5th generation BAM? Well maybe, but the use of deep analytcs and historical data has characteristics reminiscent of BI, but applied in an event driven, closed loop way. A long way from traditional BI.
