Background
When I say process, What do I mean? Everything that happens in between you ordering a book in Amazon and it being delivered to your doorstep, or You calling customer care and your problem being fixed by an agent. This series of actions is called a Business Process. A Business Process is also defined as ‘A collection of linked tasks that find their end in the delivery of a service or product to a client’ by Appian.com. These tasks that compose each business process are called events. For example, Amazon handles 1.6 million orders daily and not all orders follow the same ideal path. They tend to have variations and exceptions in the path followed, departments and resources involved. In modern enterprise systems these are maintained as detailed records of events that occur during the execution of the business processes they support. This collection of records is called an Event Log.
With this much data in hand, patterns start arising. Finding these patterns from an event log aka analyzing an event log is valuable as it helps the company to identify issues and opportunities for improvement. An approach to analyzing event logs with the help of different types of visualizations is called Process mining. These visualizations should allow process analysts and process owners to identify issues and opportunities for improvement such as bottlenecks, sources of waste, and root causes of service-level violations. And that is what Process Optimization does! A service that helps users find and fix inefficiencies and discover potential opportunities within a company’s processes to save resources (time, talent, and money).
For this report, we will be analyzing the processes of a business unit called Incident Management. Incident management is responsible for restoring services and resolving issues quickly. Keeping employees productive and happy is its main goal by ensuring that they can easily contact support to track and fix issues. Now, Let’s optimize their processes.
Materials
In order to analyze an Incident Management process using visualization, 1 Dataset: Incident Response Log, an event log of an incident management process extracted from data gathered from the audit system of an instance of the ServiceNowTM platform used by an IT company downloaded from
Incident Response Log. (n.d.). Www.kaggle.com. Retrieved October 23, 2022, from https://www.kaggle.com/datasets/vipulshinde/incident-response-log?select=Incident_response.txt
and two main software will be used, one visualization software: Tableau, and one spreadsheet software: MS Excel. The data from Incident Response Log is exported as .csv directly to Tableau, which will then be modeled into custom visualizations that benefits our use case.
Method
The focus of this report is to create appropriate visualizations that best benefit process analysts and process owners to identify, track, and resolve high‑impact incidents to optimize an incident management process, and to do that we need one main dataset, here we are using ‘Incident Response Log’.
As a process owner of Incident Management (IM), the main responsibility is Defining metrics & KPIs for their business processes to
- Create and maintain consistent processes.
- Identifying opportunities for automation.
- Perform Analysis, identify inefficiencies, and delegate analysis.
KPIs (Key performance indicators) are metrics used by businesses to help track their goals and tracking KPIs can help identify anomalies quickly which will provide a jumping-off point for larger questions.
Thus decided to create KPI Dashboard to help optimize a process where data visualizations can be used to present and interpret KPI data allowing users to quickly and easily understand trends, patterns, and relationships in the data.
The initial exploration started with 3 main questions,
- Can we create data visualization to monitor and track KPIs from the event log dataset?
- Can we map “which factor is causing the most impact”?
- Will it help process analysts to find inefficiencies & opportunities quickly?
To get a deeper understanding of how Data Visualization can be used to analyze and optimize Business Processes, I studied journal articles, conference and research papers, related to Process Optimization and found one of the main issues still encountered in the area of business process mining outlined by van der Aalst (2004a) is Visualizing Results. He states “The results of process mining may be presented in graphical form in terms of a management panel.”
Visualization is a central part of exploratory data analysis. Data analysts use visualization to examine, scrutinize, and validate their analysis before they report their findings. Decision-makers use visualization to explore and question the findings before they develop action plans. Each group of people using the data needs different graphics and visualization tools to do its work (Myatt, Glenn J., 1969)
Things to remember while using visualization to analyze Business Process mining from Making sense of data II: a practical guide to data visualization, advanced data mining methods, and applications/Glenn J. Myatt, Wayne P. Johnson.
- The focus of many data mining projects is making predictions to support decisions.
- Building models from the fewest number of independent variables is often ideal. A principal component analysis is one method to understand the contribution of a series of variables to the total variation in the data set.
Another exploration opportunity from the secondary research in using data visualization for Process optimization is the need to Improve Usability for Non-Experts: End users need to interact with the results of process mining. For this reason, hiding the sophisticated process mining algorithms, a subject comfortable only for Field experts behind user-friendly interfaces is necessary. (R’bigui, H. and Cho, C. 2017) by also improving the understandability of the results for non-experts. (R’bigui, H. and Cho, C. 2017)
Now Let us create KPI Dashboard for Incident Management process optimization: To help users better associate with the KPI dashboard, a banner element was created stating the goals and objectives of Incident Management. Step 2 is creating KPIs that are most relevant to the stated goals and objectives and how they will be measured. For incident management, the number of incidents and average time to resolve could be two beneficial KPIs. Thus the two major KPIs created were: Incidents over time and MTTR (Mean time to resolve).
Incidents over time are calculated as the [average number of incidents] within a given period.
Mean time to resolve is calculated as [Total time taken to resolve an Incident] / [# of Incidents] within a given period.
Step 3 – Now, both of these calculated metrics were graphed individually by weeks using Line Graph visualization to understand how the two metrics are performing for the specified period. Comparing both the metrics by time will allow us to closely track and identify 1) which metric is not doing well and 2) in what time frame the anomaly is happening. Based on a few user testing sessions, the KPI Line graphs created were decided to be placed next to each other in the Dashboard in order to visualize and compare the chosen data and its effect better.
Also, based on another user feedback Step 4, was making KPI Dashboard interactive to dynamically adapt itself to show only factors on the selected KPI.
“Want to look at a section – everything goes away and selected ones explode to concentrate”
-Architect, Support Process, and Strategy
The KPI Line graphs which are used to identify the metrics and time frame (problem space), were made interactive to act as a trigger to filter factors (For example assignment group) associated with the selected time frame.
By theory MTTR and Incidents over time metric is affected by two factors, 1) the number of incidents handled by assignment groups, 2) the average time taken by the assignment group to resolve an incident. Thus the assignment group will be an efficient factor to analyze and investigate in order improve the selected KPI. Step 5, Creating actions to Query data from the dataset based on user action on the KPI Line graph with respect to the factor, here assignment group, resolvers and incidents. Step 6 is creating a Heatmap visualization for each of the above to identify trends, patterns, and relationships that may be relevant to process optimization. These Heatmap visualizations were also made interactive to help users to dig in further and get clarity on who or what is causing the inefficiency.
Step 7 is using colors to helps users to establish a relationship within visualization specific to a selected KPI.
Data visualization and Insights
Process Optimization using KPI Data Visualization (KPI Dashboard)
Imagine Ben is a Process Owner of Incident management Business Unit (IM). Let’s help Ben optimize the processes of his Business Unit. Ben opens the Incident Management KPI Dashboard. He sees that a new company directive to “Increase Net Promoter Score” directly relates to Incident Management. With the goal/objective set, the dashboard now showcases Ben with two related KPI Data visualizations Incidents over time and MTTR (Mean time to resolve) Line Graph for the specified time period (March to April) to help him see outliers, underperformance, trend disruptions, etc. In doing so, it tells Ben HOW Incident Management processes are performing and points out anything unusual. The Dashboard also showcases Ben with the heatmap visualization of factors affecting KPI specific to the time frame, here Assignment groups. By showing the actual execution of the processes at this granular level, it provides insights into WHY the processes may be underperforming. Now Ben interacts with the KPI Dashboard to find opportunities to optimize his processes in order to help his Business Unit achieve their goal (Improve Net Promoter Score).
Immediately he notices that the Mean Time to Resolve (MTTR) metric is not doing so well lately. Identifies a sudden influx (increase) in the week of March 6, 2016. The MTTR has reached nearly 19.4 days from the initial 6.3 days a couple of weeks ago. There seems to be some problem causing this upsurge and when Ben interacts with it, the Dashboard dynamically adapts itself to show results specific to the week of March 6, 2016.
Thus to start with, Ben checks for independencies and sees Incidents Over Time KPI Graph also showcases a change in the week of March 6, 2016. Now that overloading might be one of the factors or start points inducing the change, the process owner starts to further investigate the Assignment groups based on their individual MTTR and Incidents over time specific to the week of March 6, 2016.
Using MTTR by Assignment Group and Incidents Over Time by Assignment Group Heatmap visualization, Ben finds that although Group 14 showcased higher MTTR i.e 80 days, the Total number of Incidents handled by them in the given week is just 2, the probability of its effect causing the inefficiency is comparatively lower than Group 70 whose Incidents total is 774 with MTTR of 17.85 days. Thus in reality optimizing recurring inefficiency (higher volume incidents) will have a better effect on achieving their goal than focusing on the anomaly that happened in mere 2 cases.
In the next step, using the sam time frame as a filter one can dig deeper into Group 70 to identify who and what is causing this high impact. Finally found that incident INC0000396 along with few other incidents handled by Responder 15 in Group 70 seems to be responsible for the major influx of MTTR on the week of March 6, 2016, using Assignment Group 70 and Assignment Group 70 > Resolved by 15 Heatmaps.
Reflections
After finding the point of the mishap and with the insights gained from the KPI Dashboard, Ben can now reach out to Group 70’s manager or the responder to further identify the cause of the mishap. Get answers to questions like “Was it the overloading? How did it affect? Is there any other factor at play? Is there a pattern?” and identify other opportunities for process improvement and implement changes as needed.
“We need to see behind the process and specifically how people interact with the pattern of work coming in.”
-Architect, Support Process and Strategy
In an overview, we found that by using KPI data visualization in this way, organizations can identify opportunities for process improvement and make informed decisions about how to optimize their processes for maximum efficiency and effectiveness. Further, continuously monitoring and tracking the KPIs to ensure that the process changes will have the desired impact on organizational performances.
The future direction of this experiment will be 1) testing if the same visualization works for other KPIs, 2) understanding the effects of hidden factors that have not been documented in the dataset and 3) Further Usability Testing.
To use a KPI dashboard effectively for process optimization, KPIs should be carefully selected to align with the goals and objectives of the organization, and they should be regularly reviewed and updated to ensure that they remain relevant and useful. In addition, it is important to establish clear targets for each KPI and to track progress toward these targets over time. As mentioned earlier by using KPIs to measure and track the performance of a process, organizations can make informed decisions about how to optimize the process and achieve their goals.
References
https://public.tableau.com/app/profile/mgkx/viz/ProcessOptimizationDashboard/KPIDashboard
https://www.kaggle.com/datasets/vipulshinde/incident-response-log
https://www.servicenow.com/products/incident-management.html#!
Myatt, G. J., & Johnson, W. P. (2009). Making Sense of Data II: A Practical Guide to Data Visualization, Advanced Data Mining Methods, and Applications (1st ed.). Wiley.
R’bigui, Hind & Cho, Chiwoon. (2017). The state-of-the-art of business process mining challenges. International Journal of Business Process Integration and Management. 8. 285. 10.1504/IJBPIM.2017.10009731.
How do Process Optimization and Performance Analytics complement each other to drive service excellence? (2021, December 20). https://www.servicenow.com/community/process-optimization-article/how-do-process-optimization-and-performance-analytics-complement/ta-p/2309413?nobounce