In a recent discussion with an ECM/BPM expert he brought up the commonly used principle that BPM can be justified by the 80/20 Pareto Principle. He suggested that it would be too expensive to define and manage a 100% of business processes so if you manage 80% then BPM is justifyable. I thought about that for a little while. The 80% rule has actually nothing to do with it. One can not invert the Pareto principle. When Italian economist Vilfredo Pareto discovered that 20% of people own 80% of all wealth he was surprised that the distribution was the same in other countries. It is probability of distribution inherent to complex adaptive systems that is most likely related to the Gauss curve. I admit that this link is intuitive off the top of my head.
I propose that this is maybe used the wrong way in BPM (and other scenarios). The Pareto Principle would for example suggest that 20% of customers produce 80% of revenue. If we now apply it the same way to BPM then 20% of processes would produce 80% of revenue? That makes a lot of sense to me. That does NOT imply that these 20% of processes have to be rigidly controlled. Are these the same 20% that make 80% of profit? Maybe there are only 20% of processes that need 80% of control effort? Possibly there even 20% of processes that cause 80% of cost? Also seems to make sense. It is most likely that these 20% groups are not related in any way. In any case it can also be 15/85 and it does not even have to add up to a 100.
Looking at 20/80 rather than 80/20 would mean that we need to monitor ALL processes to identify which 20% make 80% of revenue and which ones cost. Analysing and encoding all processes to do so is not feasable and it would anyway ruin the business. Quality, revenue and profit are a complex balance and not easy to achieve with a common approach. As I said before: In times when we look to increase business agility, it seems foolhardy to reduce people agility by insisting on vertical applications or rigid processes.
Our approach at ISIS Papyrus is to model business entities in metadata and empower users to collaborate in processes freely while enabling monitoring and auditing. The Papyrus User-Trained Agent can perform such interactive process discovery and then guide users. It is simple to add cost/time/quality/value fields to processes and fill them with federated operational business intelligence and thus identify the 20% groups. As discussed above not all processes can be monitored and tuned to the same principles.