Using statistics in BPM
1/6/2013 9:03 PM
We have seen very good and bad examples using statistic in BPM to optimize processes automatically. Some of the bad examples we have seen is a workflow process that automatically approves expenses for employees. This workflow reviews prior approval amounts to determine if the expense amount submitted by employees should be automatically approved. We thought this was a very poor example/business case for automated approval of expenses for the mere reason that most organizations have business rules around expense approvals. Most organization use a rules based approach to determine what levels of approvals are required. For example, anything =< $100 does not required any approvals, between $101 - $1000, might require 1 approval. Between $1001 – $10,000 it might require 2 approvals. Using statistics in an expense approval workflow is really not a good workflow choice.
There are many places where statistics can be used, here are some of the good examples on where to use:
Routing based on efficiency, in a support ticket application, routing the ticket for processing to the agent who’s available
Routing tickets based on who has processed the most tickets for a given ticket type
Based on a previous count of # of workflows processed, distributing the jobs evenly among employees
If a user has too many tasks to respond, then routing the new tasks to other users
Routing information based on statistics – business intelligence data
Letting users make decisions based on business intelligence data
Many BPM statistics can be computed using business intelligence data. By analyzing least/common paths of a workflow, many BI data can be retrieved. Statistics can be pulled out and injected to any workflow to control the process flow. Majority of process flows are decision-expression based. Using real-time or non-real-time statistics, the process flow can be dynamically adjusted for each executing instance.
Learn how cDevWorkflow utilizes real-time statics within the engine to optimize itself……..