On May 25 I mentioned that I would be researching (business) process discovery functionality in June and asked for input. One of the first to weigh is was Jacob Ukelson of ActionBase (whom I interviewed here back in January 2009). More on his product's capabilities in the upcoming research article but he proffers an interesting taxonomy to the process discovery task. He says:
"1. The easiest way to do process discovery would be through some automated system that looks at what people do (both on their desktop, servers, and through their communication channels) and use that to ferret out the implicit processes in their normal flow of work. That is of course (besides being extremely intrusive) impossible.
"2. The next best thing would be to use email correspondance to try and ferret out the unstructured, ad-hoc, human processes done via email and then codify the processes that were uncovered. Besides the privacy issues, doing it this way is also very difficult (if not impossible using today's technology) including issues of natural language understanding.
His comments raise a couple of disparate issues I plan to delve into more deeply as the research proceeds.
- How much of an issue is the privacy/intrusiveness aspect? Is that not cultural or geographic? At least in the U.S., I don't think employees would be surprised that the boss was analyzing what they did. But in nations that have strong workers rights' laws, maybe that's illegal?
- If you can overcome the legality/cultural thing, how much of a technical challenge is involved? What technologies specifically? When I first spoke with Jacob, the subject was how much unstructured processes were like unstructured data. I am guessing the technical challenge is similar to the years-on-end battle to improve management of unstructured data.
As I said on May 25, let me know what you think. Send me an email at dennis@ebizq.net to be included. I'll do some secondary research on your offering and meet up with you for an interview. Even if you have recently completed a survey for one of my earlier feature articles, please send me an email highlighting your process discovery features. The article on BPM Process Discovery is tentatively scheduled for release in mid June 2009 so whether you are a user or a supplier, contact me by Monday June 8, 2009.
-- Dennis Byron
Process Discovery may be technically a very difficult thing to do even when you cross other hurdles like privacy and cultural backlashes at the workplace.
Business Processes are usually run with the help of a handful of automated systems these days and it will be pretty much impossible to guess the routes a business process is taking even you track what is happening on a database. Many databases are designed to record only a modify date and time and not even what changes happened within the database. If you try screenscraping or capturing screen changes, technically it is even a tougher task.
With increasing use of business rules, it will be pretty much impossible to trace the route a business process is taking or at least guess anything meaningful out of it.
That's just my guess but companies that are actually designing these systems may have some tricks that get over the technical hump!
Regards
Nari
I totally agree Jacob Ukelson's thoughts on process discovery. From my experience it is extremely difficult to discover those informal processes. Still, there is one term I am missing in his comments: Context.
Let's assume we were able to analyse all the communication in a company. As a result we would get a very large social network, no processes. The only information you gathered was which people/organisational units need lots of communication to do their work. You will not be able to tell what the acutal processes look like.
Only if you can assign a (process)context the communication (emails) you will be able to follow the processes through the network you discovered in the first step.Of course that leads to the question who or what defines the contexts. I have no simple answer. Usually it is the scope of the project that's setting the context for me.
Another funny/interesting idea one should keep in mind when doing more traditional discovery methods. E.g. follow the documents/data and interview people is proess confabulation. (http://kswenson.wordpress.com/2008/05/25/process-confabulation/)
I hope these thoughts help your research for business process discovery. I'm already looking forward to you next article.
Peter,
I agree with you. If you take an email only route for process discovery, you will be missing the context - you'll need to try and understand the context from the email content - which is really, really hard.
That is why using a tool like ActionBase is so valuable in a discovery context - it ensures that the process context is defined by the initial email or document template that kicked off the process.