How to Analyze KPI Violations [Part 2]

This is Part 2 of the blog article on how to analyze KPI violations in Apromore. To read the Part 1, click here.

In Part 1, we performed descriptive analysis to understand the impact of conditions causing a substantial KPI violation. Here, we will focus on predictive and prescriptive analysis. Specifically, we will simulate a to-be scenario that discover the impact of changing a few parameters in the process. Let’s jump in.

Attaching a Cost Center

In Apromore, we can create simulated what-if scenarios to see the see the potential effect of changing one or more parameters in the process. For instance, we can determine the impact of increasing the number of resources from one to two or the cost implication of automating an activity. To begin, open the violation cause log in Process Discoverer. Let’s add a cost center to understand the cost of each activity. First add two new calendars — a Mon to Fri, 9 to 5 calendar and a 24/7 calendar.

Now open the cost center and add the cost per hour for each role in the process. Also attach a calendar to each role. Let’s say the “Case Officer” and “Customer Relationship Executive” work Mon to Fri, while the System work 24/7. The cost per hour of each role is indicated in the screenshot.

We can go to the cost overlay to view the costliest activity.

We see “Initial Chargeback on Mastercom” costs 27.85 AUD, the highest in the process. Perhaps, automating this process would cost lesser.

Also, we established from our bottleneck analysis in Part 1 that, “Generate Acknowledgement Email for Customer” and “Alert Merchant via Ethoca Alerts” were the major bottlenecks in the process. We can also automate these activities and determine the impact of this change to our process. Let’s now simulate the impact of automating these activities.

Define the To-Be Model Simulation Parameters

To ensure a more robust model, we can increase the arcs slider to 50. Switch to the BPMN view and save the BPMN model with simulation parameters.

Open the saved BPMN model in your Apromore Workspace.

In the Roles tab, create a new role called “Bot”. Let’s say the bot cost 10 AUD/hr can works 24/7. We can now assign the bot to the three tasks.

Also, remove the three tasks from the Case Officer assigned task.

The introduction of the bot would reduce the processing time by completing the assigned task fast. We can say the bot takes 10 seconds to complete the three tasks.

Let’s also say by virtue of standardizing these tasks with a bot, the “Send Acknowledgment Email To Customer” tasks is now done faster. We can reduce the processing time for this task from 18.327 minutes to 3 minutes.

Save this to-be model. We can now simulate this model and compare the performance to the original log.

Simulate the To-Be Model

We now have our initial model and our to-be model. Right-click each model and click Simulate. Apromore generates a simulated log for each model, saved in our workspace. Select the original and simulated log, right-click, and click Launch dashboard.

From the Simulation tab, we can see some massive improvement that the introduction of the bot will bring.

• The average case duration will change from 4.13 hours to 2.87 hours.
• The average case waiting time drops from 12.6 minutes to 4.01 minutes.
• The average case cost plummets from 101.31 AUD to 70.86 AUD.

We can also see more granular performance improvements in our to-be model.

Want to try it out?

Please find the simulated event logs and BPMN models here.

By David Obembe

David is a technical writer, passionate about making complex things simple. He also loves working with data and watches YouTube documentaries in his free time.