Modeling Processes involving Knowledge Workers

Download PDF

BPTrends, September 2013

Authors: Rick Rummler, Cherie Wilkins

For this Column we’ve decided to invite our partner, Rick Rummler, from our company, Performance Design Lab (PDL), to weigh in on the topic of modeling processes in which knowledge workers play a role. The first part of the Column was written by him. In the latter half of the Column Cherie Wilkins provides an example from our own client experience.

Modeling Processes involving Knowledge Workers

By Rick Rummler

I have tried to come to terms with the differences between our views and experiences and that of other respected experts on the modeling of processes that involve knowledge workers. I would venture to say that the majority of the business processes we address involve knowledge workers to some extent, so this is not a trivial point.

Here are a couple of examples of the views of others:

“Typical cross-functional process maps “convey that processes are predictable and sequential. We think not, especially in a services-based industry where the workforce is predominantly knowledge workers.”

-From a conversation with Janelle Hill, VP, Business Process Management Research, Gartner

“It is not easy to view knowledge work in terms of processes, because much of it involves thinking, and it is often collaborative and iterative, which makes it difficult to structure.”

Thomas H. Davenport from the Handbook on Business Process Management 1: Introduction, Methods and Information Systems; J. Vom Brocke and M. Rosemann- Editors; Springer-Verlag 2010 Germany; pg. 19.

Surely, there are differences in the characteristics of a highly regimented or automated process and those where knowledge workers play a central role. There are also important differences between manufacturing and service processes. Don’t get me wrong, l’m not saying that every process can be modeled in precisely the same way. All process models vary in subtle and important ways in order to reflect unique process characteristics and the reason for creating the model. However, in our experience there is no basis for taking a significantly different approach to modeling processes because of these differences. It is even more problematic to assume that the approach should differ based on the category of process being modeled.

Processes are most effectively thought of as a chain of accomplishments and PDL’s process definition states “a process is a construct for organizing work to achieve business value”. Taken in combination, what we are modeling is the chain of accomplishments that delivers business value. This is true regardless of the type of process and doesn’t require significant differences in one’s approach to modeling. Admittedly, effectively modeling some processes may require breaking artificial rules that are sometimes imposed on us by the process modeling software makers. In Figure 1 is a simple example of a knowledge worker- based process: the family meal planning process and tactics for conveying the characteristics of collaboration and iteration we often associate with knowledge work. (ed note: download the PDF above to view figures)

In this example we used a shared accomplishment to convey collaboration (steps #2 and #6), nested accomplishments to denote a set of accomplishments without a prescribed sequence (steps #2 and #4), and the use of a Notes band to convey supporting information without cluttering the model.

This relatively simple meal planning process illustrates the point that processes involving knowledge workers can be modeled. But even when we have created models just this simple, we have occasionally caused some stakeholders to cringe because they worry that anything like a defined step-by-step path “takes all the creativity away” or “leaves out the real essence of the work.” On the other hand, a more complex model causes other stakeholders to throw up their hands in horror because we have made things ‘way too complicated”.

But while it is tricky to figure out just what level of process detail tells the story, it can be done. Let’s look at a more complex business example from Cherie Wilkins’s experiences at an investment banking client as a way of further proving the point that mapping these processes is both possible and valuable.

The bank was very successful, but needed to grow and expand into other regions. The CEO wanted a model that was scalable and manageable. After our initial assessment, we focused first on a redesign of the Customer Acquisition and Relationship Management process. This improvement project was staffed with the executive team as a way of indoctrinating them to “process thinking.” A high percentage of this process was knowledge work. We created a cross-functional map of the process. After two weeks of analysis and “should” design, the team was able to implement several improvements addressing capacity and client satisfaction.

The heads of both the Equity and Fixed Income investment groups participated as a part of the design team. While they were quite enthusiastic about the experience and improvements, they had a very different reaction when we approached them about mapping the Securities Selection processes.

“We pick stocks. That is not a process, it is an art. It can’t be mapped. Besides, our performance is very good and we don’t need to improve it.”

Cherie asked that they indulge her in her efforts to try to map the process of Equity selection. Then we would review the results to see if it brought them any insight. Allow us to recreate a bit of what Cherie (i.e., CLW) did:

CLW: “So what is the end result of the Equity Selection process?”
Equity Director (ED): “We know what stocks we will put in our client’s portfolios.” CLW: “The sets of stocks that you will put in client portfolios — what do you call them?”
ED: “The model portfolio.”
CLW: “Okay, where does the process begin? What do you start with?” ED: “The entire known universe of equities.” CLW: “And getting from the universe to the model portfolio is all art? No science?”
ED: “Well, there are algorithms that we use for screening.”
CLW: “The analysts apply these algorithms?”
ED: “No we use a system. The database manager runs the screens against our criteria and algorithm.”
CLW: “And what is the result of that screen?
ED: “Candidates for the analysts to research.”
CLW: “So we go from the known universe to a candidate list? That sounds like a significant milestone on the way to creating the model portfolio.”
ED: “l suppose it is.”
CLW: “Could you measure the performance of the candidate list over time and get insight into the changes you make to criteria and the algorithms?” Would that be interesting to know?
ED: “Absolutely.”
CLW: “Once the analysts do their research, do they then recommend the stocks for the model portfolio?”
ED: “Yes, but I make the final selections and they don’t all get into the model.”
CLW: “Do you have a name for their recommendations?”
ED: “All of their recommendations are our Emphasis List.”
CLW: “Hmm, sounds like another good milestone. Are there any lists between that and the candidate list?”
ED: “Well the analysts each have a follow list, a smaller subset of the candidates which they are following and pulling data on.”
CLW: “So it sounds like we have a sub-process that gets us from the universe to the candidate list, another sub-process that gets us from candidates to a follow list, the next sub-process that gets us from the follow list to the emphasis list, and finally a sub-process that takes us from emphasis to model portfolio. Yes?”
ED: “Yes — that is a good way to look at it, but what happens in those “sub-processes” as you called them is mostly art.”
CLW: “Possibly, but can I go spend some time with a few analysts and the database manager?”
ED: “Sure, l’ll set that up.”

A couple of weeks later, we had ourselves a 42-step process map. We also had documented criteria for screening, standards for research, two new templates that were to be implemented (one for a detailed research report and one for communicating the model portfolio), and (trumpets sounding) a set of metrics that gave insight into all the decisions and choices that are made along the way from the universe of equities to the model portfolio — allowing for learning, management and improvement.

The director admitted that he would never have believed that this could be done and that he was pleased with the improvements. In Figure 2 is the final high level view of the process that we mapped.

So, in the case of both the Client Acquisition and Relationship Management process and the Equities Selection process, knowledge work was able to be put on paper, critiqued and enhanced. Hence, our disagreement with those who argue that “art” cannot be modeled, improved or made more transparent to managers.