Sunday, March 09, 2008

From current BPM patterns to the future patterns of BPM technology....

I read a number of recent postings on the various "BPM technology patterns": (bij Peter Fingar) and (by Jim Sinur).

(updated: Sandy - - pointed to the correct article on that I actually intented to link to. Thanks!)

Interesting typology of various BPM technology patterns - but, in my view incomplete and (a bit) outdated. STP, Workflow and Casemanagement is stuff that we understand now.
My interest is:
- What other (future) types of BPM technology patterns can we see ?
- How are the various patterns positioned?

To answer the second question, I tried to create a model, based on two dimensions.
See the following diagram:

It tries to map out BPM patterns on:
1. The process - how structured is it?
With "structured" I mean processes that we can map out, in terms of predictable events, activities and business rules. Based on incoming data (from events and activities) during the execution of the proces, certain activities are performed and decision rules determine the flow. Example: quoting a new car insurance.
With semi-structured I mean processes that are more tricky to model - some of the business rules can not be formalized and stay in the head of people as knowledge. And some of the events might not be predictable.
Example: trying to decide on a complex insurance claim
Unstructured are processes (if we can still speak of process) in which in advance very little can be said on flow, events, activities and business rules.
Example: the way I clean my house ;-)

2. Who is performing the process?
In some cases processes can be fully performed by a computer, based on predefined rules and transactions.
Other processes need human activity, based on certain knowledge and the realisation that life is more fuzzy than a computer can handle (at this stage). We can see processes in which there is most of the time one person in the lead, performing work without a lot of communication during the execution of the activity. There could be more people involved but either they are working serially or they are working on separate process parts without direct contact.
And most complex are processes where people are collaborating together at the same time, working together on performing activities, making decisions, etc.

Now, if we use these two dimensions, one can map out a number of BPM patterns...

First the well-understood patterns ...
STP - Straight Through Processing - in which a computer fully handles the execution of a process. No people involved, hands-off. Great business case for high volume, low revenue transactions.

E-Forms - this is my entry level workflow. People fill in eForms, which are send around for approval, with not much (complex) integration or complex business rules. Think - reserving parkingspaces or rooms.

Workflow - this is the typical inbox, task, form, more complex business rules and back-end integration, with some process monitoring on top. Great business case for process control, compliance, worker-support and efficiency.

Case management - a lot of discussion on the defition on this one. My view: automated process support where one can perform tasks in any order, skip tasks, add tasks. Usually quite data & document centric. Great for loosely structured processes, where we want to keep control, and support the knowledge worker with a one-stop place for all work to be done, data, documents and activity-history.

Groupware - all solutions that enable sharing information (Wiki's, email, discussion boards). Very handy, but unfortunately it creates a lot of "setup time" every time you dive into one of these components, because the process context often is missing or is not easy to establish (what't this about? what't the status? what's expected of me, now?)

Automated BPM patterns of the future

I see CEP (Complex Event Processing) as a new area, that might grow into a BPM pattern that can support complex interrelated processes. Think inter-process coordination (don't process a new policy, if we just got word that the client is in financial trouble) and more complex rule based event driven stuff (analyse this cloud of data, and determine appropriate pre-defined actions, that need to be started)

Agents and AI will provide possibilities to go even further. I am thinking of self-learning systems, that can formulate a process, based on clouds of data/events, perform it, and if needed adjust it....

Then - Collaborative BPM patterns of the future

What happens if you have a process which requires many people to work together to perform a process? Currenly we know the answer - manually planning of meetings, meeting minutes/recording of decisions, and email - lot's of email.
Typical "knowledge worker" processes are currently not well supported by BPM technology. We are far from the "high performance workplace".
Some patterns I see:

Collaborative BPM
Process engines that support working together around a structured process: automatically planning a meeting, calling a client, opening and recording a chat/IM session. All still activity focused - making sure the process can finish.
(Note: Collaborative BPM is also used as a "design time" concept, in "together defining a procesmodel")

Wilder stuff. Process engines and collaboration platforms that can handle less structured processes in the form of "processes on the fly" with "stories" that the group can create, evolve and finish. Where messaging is structured (assignments, decisions, data). See the thoughts around "Human Interaction Management".

A platform that would enable users to manage all communication around a process as well. Unified, connected to the process as context. Could probably support both CBPM and HIMS.
I see a platform in which I...
- Can manage my presence (for these types of processes, please contact me via phone of IRL unless it's weekend, than only contact me if status is RED, via SMS)
- Can route voicemails, calls, emails, documents, linked to a certain process context and activity context

It this still vague? You bet. I don't know where it's going - but I can see that the current IT support for working together is not sufficient. That troubling me. .. the growth in information overload, the growth in "setup time" to get (re)started, and the economic reality that productivity improvement in knowledgeworkers intensive processes has not significantly grown, while this area of work (services) is outgrowing manufacturing by the day...
To compete in the next 20 years, something will need to be done...

Let's check in 5 years!


Anonymous said...

There's also an article by Peter Fingar on that has a similar set of process patterns as Sinur, although Fingar breaks out the unstructured processes into ad-hoc case management and human interaction management.

Anonymous said... refers to this blog entry

Anonymous said...

Roeland, there is what I call 'Adaptive Process' that encompasses some of what you are throwing together here. You may be surprised to hear that we actually do that today. Current BPM products will not even be there in five years, because the main issue is where in the lifecycle the knowledge can be added. BPM as a concept lacks a holistic, realistic view of the business.