How to View Data Analysis Through the Lens of Process Mining

As one of the market leaders in Process Mining, the Minit team believes processes are the very pulse of an organization. The bread and butter. The essential core. But take a step back and think about a key component that those processes are built upon.

• People? Yes, the people behind processes are important.
• Mission? Yes, a mission statement can establish operational excellence as a mantra.
• Systems? Yep, the systems upon which processes run are pretty important as well.

But we’re talking about something so intrinsic to the process itself, it’s almost inseparable. We’re talking about data.

Data is the sugar to Process Mining ice cream.

Data is a fundamental component of processes and the foundation upon which Process Mining is based. Data analysis is what you do to the data to make it tell a story. It’s the shakedown that makes data reveal its secrets.

However, data analysis through the lens of Process Mining takes on a few specific elements that might otherwise be inconsequential to other types of analytics. Let’s take a look at how Process Mining views data analysis.

Prefer audio? 

Process Mining works best when you identify goals, sometimes in the form of questions, upfront. For example, when analyzing data mined from a P2P process, questions may include:

• Does my process conform to legal and company regulations? (CEO, Compliance Officer/CFO)
• Are low volume invoices taking up more than 50% of invoice approval time? (CFO)
• Does our P2P process negatively or positively impact working capital? (COO/CFO)

For analyzing OTC, questions may include:

• Why are orders from channel A processed 3x faster than orders from channel B?
• Is lower customer satisfaction in Q3 associated with delayed order fulfillment?
• How can automation help to reduce order changes?

While some data analysis will thrive on the unknown, Process Mining benefits from this type of structure. It’s less about discovering something completely unknown, and more about turning experience-based hypotheses into data-based facts.

Data Analysis for Process Mining is System Agnostic

One of the biggest advantages of Process Mining is that it doesn’t care where data comes from, it only cares that it’s data.

IT systems, as process data sources, range from big enterprise level platforms like ERP software, to mid-sized essentials like CRM, and even to rogue Excel sheets with manually entered process data from the Old School Sales Team that can't kick the habit.

Process Mining technology combs these sources to find data input for analysis, and builds a cohesive process storyboard. On the flip side, in other data analysis cases, analytics is not system agnostic and pulling data from multiple systems can lead to confusing results.

Take a lead generation campaign which advertises on social media. To analyze results, the marketer might pull data from Google Analytics for website traffic, HubSpot for lead info, and Instagram for CTR and social reach. They have to mesh the data together to form a cohesive analysis.

Process Mining, however, does the meshing for you. Being system agnostic gives this technology incredible flexibility in terms of data analysis.

Your Data Must Meet 3 Minimum Requirements

While its non-devotion to any one system makes Process Mining easy-peasy, 3 data standards by which it won’t bend, exist; Case ID, Activity and Timestamp.

For process data to be analyzed by such technology, the data input needs some conformity. Additional attributes based on defined project goals can (and should) be added before your data pull. For example, setting a date range, department or starting activity may help filter out noise, not needed for a particular analysis.

The 3 must-have data minimums include:

1. Case ID: a unique identifier that represents one execution of your process. If analyzing the order process, handling one order is one case, which has one Case ID. If analyzing the customer service process, one customer complaint is one case, which also has one Case ID.
2. Activity: process steps or status changes that occur during process completion. For example, assigning a customer complaint to a specific team member is an activity. Approving an invoice is an activity.
3. Timestamp: a marked sequence of events in time. This is especially important when analyzing activity duration.

Other data analysis activities may have looser data restrictions. In part because the analysis is not as automated as Process Mining, or because data input varies so greatly.

In processes, however, even unstructured data has the core commonality inherent to processes — start and end time, status change as a process advances, and a specific person (or bot) responsible for each activity.

Know What You’re Looking for in the Data

Let the data speak, but give it some cue cards! Digging into a Process Mining project without knowing what you’re looking for is a waste of time. Big data is, well, big. It can be messy and overwhelming if you pull the trigger without defining goals.

As data experts and our trusted partners at dab say, useful answers can only be found by asking the right questions.

"The aim of data analytics is to change data into knowledge. But marching blindly into a swamp of data without a structured approach to analysis will leave you with more questions than answers. Plus, creating the necessary data input, e.g. the event logs, correctly and selecting the right additional attributes is absolutely crucial. This is done best tool-based or by someone who understands the data sources semantics as well as the underlying data model to get the best results."

- Stefan Wenig, Managing Director at dab

Change Data Into Knowledge with Minit

Minit Process Mining, together with our data analytics partner dab, can help you change data into knowledge. Are you ready to become the process hero in your company? With our understanding of data through Process Mining, you can make it happen.

Book a product demo to get started, because we’re here to help on your data journey.

Photo by Denys Nevozhai on Unsplash