I really didn’t sit down this afternoon to write that last enormous post on the Great BPMN Debate, I remembered that Jan Recker (co-author on the research paper that sparked the debate, although not a participant in the debate) had sent me a pre-release copy of a paper that he authored, “BPMN Modeling — Who, Where, How and Why”, which summarizes the results of the survey that he conducted last year. One thought led to another, and before you know it, I’d written an essay on the most exciting thing to happen in BPM standards in ages.
Back to Jan’s paper however, which will be published this month on BPTrends. He surveyed 590 process modelers using BPMN from over 30 countries, and found some interesting results:
- BPMN usage is split approximately in half over business and IT, which is a much higher percentage of IT users that I would have guessed. Business people are using it for process documentation, improvement, business analysis and stakeholder communications, whereas IT people are using it for process simulation, service analysis and workflow engineering.
- As you might expect given that result, there’s a wide variation in the amount of BPMN used, ranging from just the core set for basic process models, to an extended set, to the full BPMN set. It would be interesting to see a correlation between this self-assessment and usage statistics based on the actual BPMN diagrams created, although as far as I know, the survey respondents didn’t submit any examples of their diagrams.
- Not surprisingly, only 13.6% received any formal BPMN training, and I believe that this is the primary reason that most people are still using only a tiny subset of the BPMN constructs in order to create what are effectively old-fashioned flowcharts rather than full BPMN diagrams.
He finished with a list of the major obstacles that the respondents reported in using BPMN, or places that they would like to see improvement:
- Support for specifying business rules, which echoes many of the other discussions that I’ve seen around having some standardization between process and rule vocabularies and modeling languages.
- Support for process decomposition, although I really didn’t follow his argument on what this means.
- Support for organizational modeling, particularly as that relates to the use of pools and lanes: sometimes, for examples, a lane indicates a role; other times, a department. There are some things happening at OMG with the Business Motivation Metamodel and Organizational Structure Metamodel that may help here.
- There are some BPMN constructs that are less often used, although it’s not clear that anyone recommended getting rid of them.
- The large number of different event types is problematic: “ease of use of process modeling is sacrificed for sheer expressive power”. This is a variation on the previous point (and on the crux of the Great BPMN Debate), indicating that actual BPMN users are a bit overwhelmed by the number of symbols.
I’ll publish a link to the paper when it appears on BPTrends; it’s fairly short and worth the read.