Evolution of Business Process Modeling

It has been a while since I wrote about the different process types, and why there is not a single silver bullet process modeling styles. If you want to read more about my first write up, you can follow this link to read more: Not all processes are created the same.

Structured Business Processes

In the early days of our business process automation space, the incumbent process modeling experience was mostly using flow diagrams. These flow models layed out a well defined set of business process steps (aka: activities) whose primary purpose was the capture best practices and standard procedures. These procedures needed to be followed in the organization that adopted them. When these structured business process models were implemented in a BPMS (Business Process Management Systems) solution, the implementing organization got a means to digitize these standard procedures and how they wanted to run their business processes and enforce execution all across. Not only business processes followed a very strict recipe, but in highly regulated industries, it offered the a way to check the audit trail records for each business process transaction, helping reduce risk and enforce compliance to policies and regulations. As structured flow diagrams became more mainstream and multiple products offered their own flavor of notation (also with their own notation interpretation), the BPMN (Business Process Modeling Notation) standard offered a common and agreed way to document business processes. This important standardization step also brought some order in the space, and helped BPMS vendors to focus efforts in supporting this business process modeling de-facto standard (or a good subset of it really!). Examples of these business process type include onboarding new customers (this is an across the board use case but very common in finance and insurance), managing approvals across different domains like order discounts and performing the multiple necessary validation steps before some coverage can be granted. Bottom line, examples exists in every industry and with varied complexities.

Unstructured Business Processes

But not all processes are the same! While structured flow-based business processes are indeed a great tool in the toolbox, this notation does not serve well other types of business processes that are completely unstructured or follow a more relaxed set of dependencies. The sequence and order in which process steps or activities are executed is determined every time there is an event in the process and this really goes against the structured and deterministic model we talked on the first section. In this unstructured modeling style, business process steps or activities are not connected via arrows or transitions. Each activity has an activation and termination expression or rule. These rules can refer to other activities or also make reference to the specific data defined for this business process (for example service type, place of origination, specific SLAs, etc). This new way of defining dependencies really allow a BPMS engine to execute following an event drive mode since each time something happens to the process (for example another step is executed), the BPMS engine needs to determine which activities need to be activated (and also likely terminated). As certain events occur for a process instance, the process has the ability to reconfigure itself and determine a new path. This is a preliminary version of the “Best Next Actions” topic. BPMS systems implementing this process modeling style have the ability to adapt and suggest what comes next based on rules defined at process modeling time. This space, is also known by the category of Case Management. When these processes heavily use human centric activities or business process steps (aka: tasks), it offers the knowledge worker to opt among multiple task alternatives. Their experience can determine which one to execute first. As these tasks are executed, the BPMS system can activate new or terminate existing tasks, helping the knowledge worker to more optimally drive completion of his/her business processes. Common examples of these business process types include investigative processes typically found in services and insurance industries. These investigative or troubleshooting processes require more flexibility (and are also more event driven) and they are better served with this alternative unstructured process modeling style. The emerging standard coming to the rescue in offering a common modeling notation (and associated semantics) is CMMN (Case Management Modeling Notation).

Wait, what about processes that do not necessarily fit either one of these models?

Alright, we now understand there are two well defined business process modeling styles. This is good. As a matter of fact really good. We are not forced to use a single notation for use cases where the notation is a stretch or not a good fit. But is it really the case that all processes fit one extreme or the other? My more than 20 years of experience in the space (and still constant conversations with prospects and customers) show that many processes fit really in the middle. The figure below is a good representation showcasing both sides of the scale and the middle ground.

Most business processes are not really all structured or unstructured. Thus, the ability for BPMS vendors to offer both process modeling alternatives and allow mixing and matching capabilities of each in the context of and end to end business process is really a BIG plus and a great blessing for its customers. In today’s world of microservices, it would not be surprising to talk about "microprocesses". This specific process modeling style really combines structured and unstructured process modeling styles, where the main process flow is modeled using an unstructured (or case management) modeling style and each business process step or activity can be delegated to a dedicated structured business process that could be maintained on its own sharing many of the good characteristics of a microservices approach. You can read a little more on this nice writing on my friend Jan Kettenis (Case Management and Micro Processes).

In this small subset of vendors that support both notations, we find Oracle. Oracle Integration Cloud allows its customers and partners to model structured and unstructured business processes. But more importantly, it allow digitizing business processes that can combine into a streamlined business process flow both structure and unstructured capabilities as outlined before. If interested, I would encourage you to give Oracle Integration Cloud Process Automation capabilities a try. You can get a free trial which you can easily start via this link: Try It!.

Well, but is there more? Indeed, yes.

Adaptive and Intelligent Business Processes

This specific business process modeling style is really an evolution of the unstructured business process modeling one (and not necessarily a whole new distinct one). I strongly believe there is still a major step to take in the unstructured business process area. And extension has to do with how unstructured business processes can take advantage of the capabilities offered by the ML (Machine Learning) and AI (Artificial Intelligence) space. While the modeling of the business process will follow likely the same notation as the unstructured world, the main difference lies on how dependencies and rules are defined. In this new world, the modeler will not need to define rules at the time of modeling the business process, but will rely on the BPMS engine to determine and predict what is the “Best Next Action”. To accomplish this, knowledge workers will initially need to execute tasks using their experience, so a history of best practices is accumulated and ML algorithms can use this wealth of data to foster recommendations and predictions on what are the most natural business process steps or activities to execute. The primary advantage of this model, is that it is not necessary to define rules as these are derived from the knowledge worker best judgement and experience. These driving recommendation forces can also be influenced over time as knowledge worker start making new decisions. While this process model style may not be attractive for all use cases, it certainly has lot of appeal in the investigative and services space. Also take in mind that only fragments of your end to end business process could use this technique. Again, the power comes in coming all these capabilities into a single streamlined end to end business process definition that you can digitize in a single BPMS solution.

Rounding up …

In summary, the Process Automation space has really been evolving in waves. This is represented by the figure below.

The first wave associated with structured business process models. This wave was followed by the unstructured wave, where it was possible to better represent scenarios that did not necessarily followed a strict path but was more event driven in nature. A middle smaller wave is the one allowing to combine these models (and strangely enough only a handful of BPMS vendors offer these capabilities) offering to model a wider variety of end to end streamlined business processes. The next big wave is indeed the one where modeling and execution take advantage of ML and AI. This is in its early stages, but it is going to be exciting. And again, it may be revitalizing to the Process Automation space.

Look forward to hearing your thoughts and comments if you want to leave one below!