I just love that term: “Process Confabulation“. It sounds like something that WC Fields or Mark Twain might say. I saw it used in a slide share presentation from Michael zur Muehlen. What does it mean? It refers to an interesting problem in uncovering the process that a business organization is currently doing. Before any BPM project, you must first answer the question: “What is the current business process?” The first place you will look is in the organizations policy manuals. If such manuals exist, they are very likely to be out of date. Next you interview the people involved.  The catch is that much of their knowledge about the way they work is tacit knowledge: things they know but they can’t really explain to others. Yet if you ask a professional how they do their work, they can’t say “I don’t know”.  How do they justify their salary if they don’t know what they do?  They simply can not admit this.

So what do they do?  They confabulate: they make something up. They think carefully about what they think they should be doing. They then describe the business process which is somewhat idealized. It may be the processes that they think they do when they get the chance to do things right.  They may not actually be aware that they are making it up; it is a trick that the mind plays on us.  Because they are performing the job, it is inconceivable that they don’t know how to do the job, and as a result they believe whatever explanation they come up with.

Fujitsu has pioneered a new process discovery service to help avoid Process Confabulation. This process discovery service works by extracting data from log files that you already have in your organization. (See EBizQ, TMCnet, onSTRATEGIES.)  Think that you don’t have any such log files? Think again.  Almost every step of an important business process involves some sort of record keeping.  Currently those separate applications are not linked together in any way, but it is possible to take dumps of these database tables, find correlation values between them, and then produce diagrams of the business process.  As a service, Fujitsu can discover these processes in about a week or two.  The interesting thing about the diagrams produced is that they a factual representations about what has happened in the past. They are not confabulations. They serve as excellent reminders of the current process, which then those within the process can use as a starting point for talking about the process with those people who participate in the process. The key to the Fujitsu approach is that there is no need to put a sensor into your environment ahead of time: you can start immediately with the data that you already have.

Vance McCarthy from IDN pointed out to me this week that this new capability may have an interesting dynamic in the way that BPM is implemented. BPM project planning is done today with IT and business people together, but it is the business people in the driver’s seat defining the process, and the IT people simply implementing. Process discovery from data log files allows the IT people to show the business people the actual existing process. The discovered processes are credible, shifting the roles that the IT folks can play. It is hard for me to guess what effect this shift might have, but it does remind me of a trend that was discussed by Shoshanna Zuboff in her book “In the Age of the Smart Machine” about the transformation that occurred when retail grocery stores got barcode scanners; they were able to use the additional information about sales patterns in order to dictate to suppliers what they wanted, instead of the other way around. It is possible that process discovery from log data may be an empowerment to anyone who masters the technique quickly.

There is intense interest in automated process discovery.  These days I am getting so many queries that I may devote a few more posts in the future about some of the less obvious aspects of this developing field. I will try to use the term “Process Confabulation” as well — simply because it just sounds so unusual.