Analyzing Process Flows

In SAP Signavio Process Intelligence, every analysis is built by using analytical widgets. This lesson focuses on the two widgets that visualize process flows (variants) by backtracking the executed events.

Every analysis starts with the most important part, the discovery of a process! Within SAP Signavio Process Intelligence, the Process Discovery Widget focuses on visualizing the different executed flows (variants) of a process and providing an overview of the actual AS-IS state.

Select the play button to watch the video on the Process Discovery Widget.

The first widget of every investigation provides a view of the actual process flow from the standpoint of events. It can also be a good entry point go get an initial understanding of the sequence of events.

Let's look at the image below of a process flow. Explore this main process flow in the Process discovery and try to answer the following question.

Let's look at the image below of a process flow or explore this main process flow in the Process discovery by adding some activities and sequences. Then, try to answer the following questions:

• How many T-Shirts were delivered in total?
• How often did the shipments fail? What kind of shipment failed more, express or standard shipments?
• What is the least common activity?
• Do you already notice some problems in the process flow?

Apparently, not everything went as smooth as intended.

We see that in total 805 T-Shirts were delivered.

We found that out of 551 standard shipments 106 of them failed and that 55 express shipments (out of 415) failed. Which indicates that standard shipments go wrong more often.

The activity order canceled counts only 72 events and therefore is the least common in our data set.

Things did not go as planned. We see that goods are already shipped before the payment is done and that orders are canceled after the printing.

Each different flow of events can be considered a process variant. The variant explorer, lists them up and provides insights to the most common/most expensive variant. 

While creating this widget, you can decide between with conformance or without conformance to filter the result, but this requires a linked BPMN to your data (we explain how to do this in the Process Conformance section).

Select each button for more details.

To investigate the process flow based on variants, the variant explorer widget would be the right choice!

• Reopen you investigation.
• Create a variant explorer widget (Make sure it's set to Occurences and Case) and answer the following questions:
  • What is the most common variant? Describe the process flow.
  • How many percent of cases belong to the most common variant? Tip: Change the settings from Number to Ratio
  • Among the top 3 variants what is the process step with the highest duration? Can you make a assumption, why it takes so long? Tip: You have to set it to Cycle time

We see that the most common variant includes the standard shipment and 228 (26%) of all cases in the data set. It looks as follows: Start → Receive Customer Order → Receive Payment → Ship Goods Standard → Receive Delivery Confirmation → End

We also spotted that the time between Ship Goods Standard  and Receive Delivery Confirmation takes at longest (12 days), considering all cases included in the top 3 variants. A possible explanation could be that the first shipment fails, and a looping in this activity occurs which makes it take so long.

The entry point for getting an initial understanding of sequence of events. It provides a view of the actual process flow from the standpoint of events.

A process can take different paths. The variant explorer lets you investigate them and find interesting behaviors.

We already had a look at Process Discovery and the Variance widget. Filters are a great way to dive deeper into investigations. For example, could we filter for just the cases where a shipment faults or investigate only the cases with high order amounts? Since filters apply to every widget they can also be used with all other widgets. Let's have a look at them.

Filters can be used to reduce the data set and focus on certain criteria. They help create a refined view of your process data and focus on aspects you're most interested in. Here are some key points about filters:

• Filters are additive.
• Filters can be applied on the investigation, chapter, or even widget level.
• Filters add or remove cases from the data.
• Filtering impacts the widgets in the investigation.
• For example, I want to filter cases for "premium" customers on a chapter. 

Watch the video below on how to apply simple filters on chapter level.

As you have learned, each chapter represents typically an analysis with a certain focus and filters can be used to reduce the data set and focus on cases fulfilling certain criteria.

• Reopen you own investigation. Create a chapter "analyze customers" and set the chapter filter to premium customers.
• Answer the following questions:
  • How many orders of premium customers did we receive? Tip: Use the process discovery!
  • What is the most common flow of events? Tip: Use the variant explorer!
• Apply a additional filter for T-shirts with printand answer the first question again.

You used simple filters to further investigate you data based on certain conditions. By using chapter level filters your filtered results applied to all widgets in the same chapter.

By applying filters we found that out of all orders received (877), 320 of them are from premium customers. Further, 154 T-shirt with print orders came from premium customers.

For premium customers, the most common process flow is described as follows: Receive Customer Order → Receive Payment → Ship Goods Express → Receive Delivery Confirmation