Last time we did an introduction of the Power Automate Process Mining. In this blog and the next one we will get into the knitty-gritty of process mining. Our focus today will be on the cloud version of Process Mining and the next time we will dig into the Desktop version.
At my company, we have a long history of employees not submitting their timesheets on time, and I have been working with my company to mine into the timesheet submission process and get some insights into how the process actually works. So, today we will use some real data, a CSV extract of our Timesheet Submission process for this blog to go through the steps required for ingesting data into the Process Mining tool and then analyzing the insights generated. It is key to note that just like PowerApps and Power Automate, Process Mining has many many connectors available as well, so feel free to create live connections to your system so you can monitor as your process evolves.
1. Navigate to https://make.powerautomate.com
2. Ensure you are in the correct environment and select Process mining on the left menu.
3. Select Start Here
4. Give your Process a name and a description and hit Continue.
5. Upload your file or use a connector to connect to your system and select Next.
6. You will be directed to a screen where you will have the ability to preview your data.
7. If all is looking good, click Next and you will be directed to the data transformation screen. This is where its important to ensure that your data types are correct i.e. Date/time or decimal fields should not be text fields.
You can simply right click on the column --> Change type --> select the correct data type --> repeat for other columns.
Once completed, select Next
8. You should now be directed to the mapping screen. This is the most important part. If you remember, an events log requires three critical attributes and, on this screen, we will have to map our columns to these three attributes types:
Case ID: The unique identifier of the event taking place.
Activity: What is happening in the event
Timestamp: When is the activity happening
If you have additional columns, you can map them to the following types:
Case level attribute: These attributes are uniform across the entire case i.e. Region, Project Name, Department
Event level attribute: these attributes are specific to the activity types i.e. Approved On, Submitted On
Financial per Event: These are the currency columns.
9. Click Save and analyze.
10. You should now see your process map generated.
Across the top are some KPIs such as:
Average case duration: How long does it take on average to complete the timesheet process.
Self-loop cases: How many times is an activity directly followed by the same activity.
Loop cases: How many times an activity is followed by the same activity within a series of steps, but not directly.
Rework cases: Number of loops and self-loops
To the right of the process map, you will see.
The number of variants we are working with: How many ways is this process getting completed.
Number of cases
Number activities
You can click on the Map tab to get a larger view of the process map.
You can see the frequency of each of the variants. We notice that the most frequent variant is Draft --> Submitted --> Approved.
If we look at the map by performance, we notice right away that it takes a long time for drafts to move to submitted, which immediately makes me think that may be employees are forgetting to hit the submit button.
If we want to look at the average duration for each activity, we can select Mean duration in the top dropdown. We can see that when recall requests are done, the process takes longer.
You can also look at the Reworks. We can see 465 cases self-looped on the draft status meaning the draft status was followed by the draft status. Very likely that people were just making changes to their entries.
You can go to the Variant DNA and see all the different ways your process is getting completed. This allows you to easily visualize your processes in a way that displays the events that belong to each variant using blocks that look like DNA.
I see in my case that there are 27 variants in my process and Draft to Submitted to Approved is my no 1 and most frequent variant with 7084 cases in it.
The ones at the bottom are the least frequent.
The process map, just like a powerbi report/dashboard, has drilling in capabilities. If we go back to the Summary tab, you can click on your frequency bars and watch your process map change. I clicked on my variant 1 and saw my process map change to a clean, straight line and beautiful process.
I also get other insights such that the average duration for that variant is 7.96days, there are no self-loops or loops happening and there are 7084 cases following that path.
I tried this out with another variant and there seems to be a deviation before it gets approved.
Lastly, I just want to touch on Copilot in the power automate process mining tool. It has the ability to provide process insights through quick and easy natural language expression. It can give you insights on your process and recommend solutions in power automate process mining.
Use the suggestions provided by copilot to get insights on your process. You can also ask your own questions by using the text field. It is important to note that the Copilot in Process Mining is still in its baby stages, so it may not always be able to smartly answer your questions, but it is getting there.
That’s all for today but stay tuned for my next blog on Power Automate Process Mining on the Desktop.
See you next time!
About Me
I'm Leoza Kabir Barker, a Functional Consultant at XRM Vision with a focus on the Power Platform. Through my expertise, I aim to streamline processes, optimize operations, and maximize productivity.
Connect with Me
Blog: www.leozakabirbarker.com
Comments