It is not a stretch to see how process discovery leads to precision training opportunities. Precision training is akin to precision medicine. While broad-spectrum drugs have its place, precision medicine has revolutionized how ailments are targeted and treated – specific dose aimed at precise targets at scheduled times. Precision medicine increases efficacy and reduces side effects compared to broad-spectrum medicine.
Similarly, generic training has its place, but the peanut butter spread approach diminishes the ROI. Instead, precision training will increase efficacy, reduce cost, time, and effort and allow an employee not to feel stuck in an irrelevant training program.
However, delivering precision training has not been possible due to a lack of observability and identification of the unique learning and development needs of individual employees. Even if they did, it was not possible to observe and document training needs at scale for a specific individual at an elemental level - mainly when there are thousands of employees. Until now.
With the advent of cognitive technologies such as computer vision powering process discovery, it is possible to pinpoint where employees are having difficulties and identify nuggets of training that are precise and effective. That is what we at Skan.AI solve for – observing every digital trace of human work and systems interactions and create a holistic picture of a process with evidence-based information that provides insight into time taken by individual associates at each step, a comparison with peer group, and normalization across several parameters to deduce a precise picture to pinpoint training needs.
Without evidence-based training needs assessment at an individual employee level, the corporate learning and development teams in large enterprises designed training programs that are often broad and general as they lacked the understanding of the specific needs of organizations or individuals. Instead of the peanut butter training methods, what if one can identify the requirements that are evidence-based and pinpoint the needs of individuals or groups based on their work patterns? It is not a pipe dream. New generation process discovery can pinpoint the needs of groups as well as individuals to tailor training or just-in-time, on-the-spot tutorials to help associates overcome the challenges. Here’s how!
In recent years, techniques such as Process mining and process discovery are gaining traction as a way to unveil the invisible enterprise and unlock the value from understanding the hidden processes and numerous pathways. Process discovery leads to many positive outcomes, including mitigating process leakage, aiding the hardening of the automation efforts to cover the full spectrum of process nuances and achieve process compliance.
There are two main techniques to discover process variations – the traditional method of application event log analysis and the next generation techniques using computer vision and machine intelligence.
The concept behind computer vision is an unobtrusive observation of every human interaction with systems and compiling a process metamodel from the digital footprints of all activity. (Mind you, the computer vision is not the same as recording using your webcam and focuses primarily on what is on the screen and not what is in front of it.)
However, while using computer vision for process discovery, an often overlooked benefit is the ability for process mining and process discovery to identify learning and development opportunities that are precise and pinpointed. Let’s delve into how.
Let’s assume the process discovery involves 100 agents who in the accounts payable of a large multinational firm spread across three locations.
A typical process discovery endeavor will involve placing unobtrusive probes (or VPA - virtual process analysts) on the terminals of agents who are a part of the study. Then the VPAs record every human and system interaction and capture images of such events. Through advanced image analysis and human input to tag events into activities, a process model emerges – with a few standard variants that are necessary to accommodate specific workflows and decision points and several other process variants which are drifts from the expected path.
So why do these process drifts occur? There are many reasons, but chief among them is the tribal knowledge in staff that goes beyond the standard operations manual, the complexity of specific cases, the systems limitations, the established practices within a team or individuals, and gaps in expertise.
Let’s focus on patterns that emerge from process discovery and analysis that can lead to precision learning and development opportunities as well as work reallocation.
For example, let’s assume for activity 8, the average time of completion is 12 minutes, but 4 of the 20 team members in that group are consistently beyond the expected norm – say 18-22 minutes. If this pattern is consistent, after accounting for things like breaks et al., it is evident that the four-team members are struggling with activity 12. An analysis of the process steps and actions may reveal that for that step 16 of the 20 associates are using Excel, particularly a pivot table, whereas another team is working off a long table scrolling up and down to find the specific item. Viola! After talking to the team, it becomes clear that the four team members who are taking twice as long are not familiar with how to build and use a pivot table. A training of one-hour will resolve the issue and help boost productivity. In addition to productivity, the precision training will help raise the morale of the four-team members who were struggling and felt they are underachieving when compared to their peers.
Let’s assume, the team in Mexico is taking 30-minutes per case, whereas Ireland is at 32 minutes and the Philippines is lagging at 44 minutes. An examination of the process may reveal that the teams in Ireland and Mexico had an opportunity to complete comprehensive training on the new system upgrade, whereas the Philippines team has yet to complete the same exercise. The lack of training on the newer version of the software could be easily remedied to put the Philippines team on par with the rest of their peers in other countries.
Now, let’s look at how even nuanced behavioral patterns can be identified through a more in-depth process analysis.
Let’s assume one of the claims processing agents within a group of 10 is consistently raising exceptions from the claims cases. Of course, it is possible that this agent is getting many complex cases or over-indexing on fraudulent claims. However, that notion is simplistic. Accounting for variations in case complexity and fraudulent claims, it may still be that the agent is very high on exceptions and rejections and upon more in-depth examination it seems like those cases are representative and not outliers warranting such high rates of rejections and exceptions. Discussing with the agent and observing the patterns, one may deduce that the agent felt it is their job to lower or minimize the claims outflows as it is “money out” of company coffers. A behavioral training that claims are a part of doing business and customer satisfaction during claims has a direct bearing on reputation and renewals would help the agent mend his/her ways.
Besides, evidence-based process documentation will help teams train the workforce on real, rather than assumed processes. The reality-based process documentation is also an invaluable input to the groups that are trying to automate, transform, or outsource the operations.
Many of these examples would not have been possible without the power of observation, recording, and synthesis of digital traces of human work into an actionable process model.
We at Skan are excited about the possibilities and how the power of process discovery leads to precision training opportunities. If you wish to find out more about how you can harness the power of cognitive process extraction, please get in touch with us.
Satya Iluri is Vice President of Strategy, Marketing, and Customer Success at Skan, an AI-powered process mining and process discovery platform.
“One of the things you don’t ever want to do is to automate a bad process. You are just going to make bad things happen faster, and that is not what anyone wants.”