Strategic Insights from Enterprise Operations

Skan Co-founder and CPO, Manish Garg unveils the importance of computer-vision based process observation as opposed to traditional methods of manual observation. “The modern method of computer-aided process discovery—the art and science of observing at scale, covering diverse geographies, teams and case logs—all done unobtrusively, is a game changer.” 

"To acquire knowledge, one must study; but to acquire wisdom, one must observe." —Marilyn vos Savant

From the time of Frederick Winslow Taylor and the advent of scientific management, companies focused on standardization and automation (or mechanization) as the corporate nirvana. Don't get me wrong—they are worthy goals and have a time and place in operational optimization. However, there are many other considerations to satisfy the multistakeholder ecosystem in addition to worker productivity and throughput.

Indeed, Taylorists measured and observed workers doing their tasks to calculate and analyze to glean ways to eliminate waste in any process. But today, thanks to the advent of modern technologies such as computer vision and machine intelligence, observation at a vast scale and high precision is possible.

Manually observing or shadowing a worker or two working on a dozen cases is exciting, but it is just the tip of the iceberg. The sample size of what a couple of consultants observe and report does not extend to tens of thousands or sometimes millions of actions, interactions and transactions in a large corporation. Furthermore, when someone is watching, the subjects' behavior (the employees) will inevitably vary as they tend to be on their best behavior and follow the book. In addition, asking someone how they do their work does not always capture the subtleties and nuances, shortcuts and friction points, rework, and bottlenecks.

The discovery of the volume, velocity, variety and variability in processes is possible when one observes a large group of employees over an extended period. Hence, hiring consultants does not always translate into an accurate picture.

The modern method of computer-aided process discovery—the art and science of observing at scale, covering diverse geographies, teams and case logs—all done unobtrusively, is a game changer. Typically, most process discovery tools rely on a small probe that observes work in real time as knowledge workers interact with digital systems. 

This process genome mapping provides visual evidence of how work happens as it happens. This treasure trove of process data can lead to strategic decisions and improvement interventions beyond conventional logic and cookie-cutter solutions.

In our work with several global corporations, here are some things the observed data has unveiled:

• Golden process: Companies develop standard operating procedures, operational manuals and checklists, often focused on a single golden process. However, process intelligence often reveals that a golden process is a myth. Instead, there are several possible permutations and combinations based on case complexity, the importance of the customer, the transaction value and other factors.
  
  
• Nonstandard processes lead to delight and NPS: Standardization, mechanization and automation are a spectrum of solutions to address process improvements. Sometimes, to satisfy the customer (or other stakeholders like suppliers, the regulators, employees, distributors, et al.), nonstandard, non-automated workarounds are critical. They avoid reverting to the mean and becoming the lowest common denominator.
  
  
• Not every process is a nail: In recent years, companies have licensed RPA (robotic process automation) platforms and have tried to use this hammer indiscriminately across processes that are not pliable to automation. This has led to a lot of heartache and disillusionment since the automation failed often and could not scale.
  
  
• Just-in-time precision training nudges: General-purpose training helps provide the foundation but lacks the precision and specificity as it applies to a specific individual and the type of case they are working with. Process intelligence algorithms can pinpoint the areas where an employee may require help and offer it seamlessly. This reduces the supervisory burden and the number of escalations. These machine learning models can be the foundation for an AI supervisory.
  
  
• Intelligent capacity planning and case allocation: After analyzing a sizeable transaction volume and process runs, the process intelligence algorithms can identify which set of cases to route to which teams (and individuals) and the caseload based on complexity and other metrics. (So, it will no longer be a situation where 100 cases are divided into 10 cases each among 10 process workers.) Furthermore, volumetric forecasting can help provide a blueprint for capacity allocation based on incoming caseload forecast, the variety and variability of cases, and the relative complexity of the case log.
  
  
• Interviews do not tell the whole story: Whenever someone is asked about something, what they tell is superfluous and does not cover all the nuances. But more importantly, interviews do not elicit the basic dimension of knowledge.

Of course, whenever someone mentions observing work, questions around personal privacy and information security come to the forefront. And of course, these are important considerations, but sophisticated process intelligence platforms protect information and privacy through various measures, including inclusion and exclusion lists of applications, selective redaction of information, anonymization and encryption, and retaining confidential information within the firewalls of the company.

The art and science of process discovery and intelligence are exciting new developments to understand the nature of work patterns drawn from a statistically valid sample size. In addition, the observed process data add visual evidence to the expertise of process owners/consultants and operational executives to make informed and intelligent decisions.

Read the original Article on Forbes.