dangrady510
ServiceNow Employee
ServiceNow Employee

How to use Process Mining to identify automation opportunities

 

This post is part of the Process Mining Use Case Series where we’ll focus on different techniques to identify process inefficiencies, non-conformant activities, and improvement opportunities.  These posts will be broken into two sections – how to do the analysis and how to configure the Process Mining project.

 

Organizations are constantly looking to automate manual tasks to reduce operational costs, improve efficiency, and free up employees to work on strategic, or high value activities.

 

Process Mining can be a valuable solution in this pursuit by helping to identify automation opportunities.

 

We can apply Process Mining across IT, Customer Service, HR Service Delivery (or any other workflow for that matter) and look for tasks that take anywhere between 2 and 30 minutes to complete.  These tasks are typically routing or rerouting scenarios, routine approvals, requests for information, or data entry type activities and are ideal candidates for some form of automation.   The 2 to 30 minute time window can certainly be adjusted but the point is we can use Process Mining to focus on work that hasn’t already been automated (less than 2 minutes to complete) but doesn’t take significant human thought or involvement (less than 30 minutes).

 

This focused analysis will help us pinpoint automation opportunities that will speed up service delivery and increase overall productivity.

 

 

How to do the analysis

 

This approach can be applied to any workflow(Incident, HR Case, Customer Service Case, etc).

 

The only specific requirement from a configuration perspective is you should include State (or some field that indicates the start and end of work that is being done) in your Process Mining project.

 

Open your project to the Analyst Workbench view

dangrady510_0-1727810137110.png

 

 

From here we are going to use the “Transitions” option from the Advanced filters section in the lower left hand corner of the screen.

 

dangrady510_1-1727810137112.png

 

Click on the Transitions button.

 

The Transitions filtering option allows us to isolate specific Activity combinations like work moving from the State of In Progress and then being directly followed by the State of Resolved or work assigned to Group A and then eventually followed by Group B. 

 

We can also use Transition filters to apply additional constraints like show me all the work that took longer than 5 days to move from Group A to Group B.

 

So where traditional reporting and its condition builder only allow you to focus on the current state of the records, Transition filter will allow use activities that happen during the process to isolate and analyze the data.

 

In the Transition filter dialogue enter of your first condition of State is “In Progress” and set the Occurrence is value to First.  (this tells the engine to look for the first time the record moved into the In Progress state only).

 

dangrady510_2-1727810137116.png

 

Click on the Add next activity button.

 

dangrady510_3-1727810137117.png

 

Leave the relationship as “Directly followed by”.  We want to focus on work that is easily completed.  If this work is moving to multiple states before it is resolved or completed that lessens the likelihood that it would be good for automation.

 

dangrady510_4-1727810137118.png

 

For our second condition enter State is “Resolved” and set the Occurrence is value to Last.

 

dangrady510_5-1727810137126.png

 

 

*Note:  Depending on workflow and your configurations you will likely have different field names and values.  We are just trying to illustrate the concept here.

 

So we have our criteria set now we need to add the most important piece for our analysis.  The time constraint.  You should adjust this for your world but for our example we are going to choose 2 minutes as our starting point and 30 minutes as our ending point.

 

Why start at 2 minutes?  Because if something is moving from the State of In Progress to the State of Resolved in less than 2 minutes it is likely already automated.  If you don’t think that is the case in your organization adjust accordingly.  On the flip side if you think 30 minutes is two long feel free to shrink that down to work taking less than 15 or 10 minutes.

 

Click on the Add Constraint button on the left hand side of the dialogue.

 

dangrady510_6-1727810137129.png

 

 

 In the name field enter “Longer than 2 and less than 30 minutes”.

 

Set the Start value to 1 (meaning the first condition in the chain).

 

Set the End value to 2 (meaning the second condition in the chain).

 

Set your Min duration to 2 minutes. 

 

Set you Max duration to 30 minutes.

dangrady510_7-1727810137136.png

 

 

Click on the Add constraint button.

 

Now click on Apply in the lower right hand corner.

 

Click on the arrow to the right of the Related List Condition text to expand the menu.

 

You will likely get a pop up indicating that your Task is running in the background.   This is an indication that a mining job as started.

 

dangrady510_8-1727810137142.png

 

 

Click on the Close button.

 

Once you receive the indication that your mining job is complete.

 

dangrady510_9-1727810137146.png

Open the Scheduled tasks panel on the right hand side of the Analyst Workbench to pick up your completed tasks.

 

dangrady510_0-1727810736313.png

 

 

Click on the View result button for your task.

 

dangrady510_1-1727810736336.png

 

 

In my case, I’ve got 838 incidents that on average took 11 minutes to move from In Progress to Resolved.  Looking at the histogram there is a big chunk that took between 3 and 6 minutes and 6 and 9 minutes.  There has to be some automation opportunities in there.

 

It’s at this point you can use other Process Mining capabilities to continue to the conversation with this data to identify the specific opportunities.  Maybe you look at these 838 records or you run Clustering or RCA on this subset using the investigation options. 

dangrady510_2-1727810770344.png

 

Maybe you use your Breakdowns to further segment the data.

dangrady510_3-1727810770354.png

 

 

There are so many options to continue digging.

 

The point of this Process Mining Use Case Series post was to give you the initial inspiration for where to start looking for automation opportunities.

 

Automation opportunities as an Improvement Opportunity

 

In addition to looking for automation opportunities in an ad hoc fashion like we did above.  You could also use the same technique and create a rule based Improvement Opportunity to be surfaced on the Summary and Insights page of the project.

 

To do this edit any Process Mining project.

 

Go to the Select improvement opportunities section.

 

dangrady510_4-1727810808896.png

 

 

Click on the New button.

dangrady510_5-1727810808898.png

 

 

From the Improvement opportunity definition page click on the create button for a Rule-based opportunity.

 

dangrady510_6-1727810808916.png

 

 

In the Create Improvement opportunity definition dialogue enter this Message – “Automation opportunities: Longer than 2 minutes and less than 30 minutes”.  Select a Category of Automation.  If you have a Performance Analytics KPI that would be impacted by increased automation add that to your Impacted KPIs section.

 

dangrady510_7-1727810808929.png

 

 

Click the Configure button.

 

Configure the Improvement Opportunity exactly as we did for the Transition filter above.

 

dangrady510_8-1727810808949.png

 

 

The next time you mine this project you will have a new entry on the Summary and Insights tab that highlights these Automation opportunities to jump start your analysis.

 

dangrady510_9-1727810808980.png

 

If you are looking for more in-depth training you can use the Process Mining Academy library of content

 

You can find other Process Mining use cases here