“To acquire knowledge, one must study; but to acquire wisdom, one must observe.”
― Marilyn vos Savant
Observing how people actually do their jobs, rather than relying on them to describe their work, is a better way to understand business processes. Interviews and surveys miss out on important details, such as unconscious knowledge and variations in how people perform tasks. A better method is the usage computer vision to watch users' screens and track their interactions with digital systems. This creates a data-driven process map that is more accurate and complete than traditional methods.
Observation: The path to wisdom
Observation is the way a child learns and develops. An infant does not read books but observes and mimics the behavior of those around them.
In science, the role of observation is undeniable, and it is the cornerstone of scientific advances. Scientists use observation as a method for compiling data, which enables them to construct and then test hypotheses and theories. Scientists observe in many ways – with their own senses or with tools such as microscopes, scanners, or transmitters to extend their vision or hearing.
Frederick Taylor and his scientific management techniques have introduced the science of observation to the field of business and management by way of time studies. (Taylor’s techniques combined with Frank and Lillian Gilbreth have become the famous Time and Motion Study.”
Taylor’s time study is the observation of a task – continuously and preferably unobtrusively - using a timekeeping device) to monitor and record the time taken to accomplish a task and it is often used when
- The tasks involved are repetitive
- Varying cycle times
- Involves variety and dissimilarity
Why observation is a superior method for business process analysis over other techniques such as interrogation and interviews or self-reporting? The reason interviews fall short has to do with the inability of the analyst to grasp every nuance and document it into a process. More importantly, it is the tacit dimension of knowledge – where people know more than they can tell, and the divergence between stated and revealed preferences.
For example, if someone were to ask you “How do you drive?” – you cannot describe in detail the extensive and intricate details that go into driving and events that happen in microseconds in real-time and impulsive responses thereof. This is an example of the tacit dimension of knowledge.
And then ask someone about their risk tolerance in investing, and they may say they are “conservative,” but the preponderance of short positions and sales of naked calls describe a different reality. Or they may say they are aggressive, but yet their portfolio may show 80% in cash and money market instruments.
Or bringing it close to home, ask someone how they eat, and the answer could be “I eat healthily,” but the grocery bill laden with Cheetos, Tater Tots, hamburgers, and corn dogs may tell a different story.
The examples mentioned above portray the differences between stated preferences versus observed preferences.
So, observation is woven into the fabric of our life, science, and business.
Observation is more than seeing things in plain sight. Observation is a synthesis of sight, perception, context, prior knowledge, and significance.
Modern business analysis and process mapping also follow the observation method – typically termed “Shadowing.” While the technique itself is appropriate, the accuracy and the completeness are questionable, given the vastness of today’s enterprise processes and the intricacy of completing tasks across multiple applications. Furthermore, traditional process mapping is limited to a happy path (or a golden reference process) and one or two exception paths, while the reality is each process has dozens of variations.
To avoid the incompleteness of process information, process mining, which derives a process map by analyzing application event logs, has become a viable and valuable method. Alas, the challenges with this approach include:
- Not every application writes event logs
- A lot happens between committed states of data – a copy/paste from an email, a calculation on Excel, visiting an external database for some information, et al.
- And green screen apps and virtualized systems galore.
To solve these problems and enable observation at vast scale and high precision, Skan uses computer vision to record process participants’ essential screen interactions with digital systems, and then using machine learning and data science derives a process metamodel including all the variants of work.
All this happens without any systems integration or access to log files. Furthermore, personal privacy and information security are embedded into Skan’s cognitive process discovery platform with techniques such as redaction at source, and inclusion/exclusion list of applications.
The output from Skan is visual evidence and is data-driven, making it a treasure trove of process data to analyze, synthesize leading to automation, transformation, optimization, and conformance.
Contact Skan to learn more about AI-enabled Process Discovery can help your enterprise unveil the invisible enterprise.
