Process Mining: A Data-Driven Approach to Business Improvement.

Are you an analyst, product or program manager looking for new ways to improve business processes and outcomes?

Yes!

Then this overview of process mining is for you.

In this post, you'll understand:

• What process mining is

• How it differs from traditional process mapping

• The benefits and challenges of implementing process mining

• Real-world industry use cases and applications

• How to get started with process mining in your organisation

Wrapping up with actionable steps, further reading and additional learning resources.

Definitions

Okay, I know, you know what a process is, but lets formalise it:

What’s process mining?

The Differences

Process Mapping

Process mapping can range from simple to complex, depending on the target problem.

Examples of processes within a retail environment include:

1. Setting up a Product SKU 

2. Creating an Order

3. Supply Chain Management 

It starts with the “as-is” i.e. what’s the current process?

A time frame is normally agreed for the analysis, before talking with stakeholders to understand their process activities, needs, challenges and frustrations.

Depending on complexity, you may need Visio or other alternatives to document the process. Once documented, the process map is reviewed and approved by stakeholders.

The inherent dangers with “manual” process mapping are:

• Stakeholders may have no / incomplete process knowledge

• Not all activities are captured

• Gathered information is inaccurate

• Gathered information is not evidence based - perceptions

 Note: If the “as-is” process can’t be defined and validated, any “to-be” improvement is compromised.

Process Mining

Inspired by IBM’s definition, the process mining funnel is illustrated below.

The 3 retail examples I provided earlier all require IT applications and systems.

Process mining involves gathering telemetry data (e.g. event logs), from these systems to detail the “as-is”. Algorithms are then applied to understand trends, before workflows can be validated and improved - “to-be”.

The key difference between process mapping and mining, is the evidence based approach, i.e. it can be proved. Manual process mapping can also be evidence based, however, with telemetry data, actual usage / process participation is digitally detailed.

Benefits

Cost Savings

One of the main benefits is costs savings, see how Aldi saved €3.1 million in value realised and costs avoided. 

Evidence-Based

Process mining shows what’s actually happening within an organisation, rather than perceptions and assumptions.

As an illustration, if stakeholders state that “sku setup” takes 2 days to complete whilst the telemetry data from IT systems shows 4 days, the digital evidence outweighs stakeholders view - action to analysis the discrepancy.

NB. The figures in the table are for illustration only.

Increase Productivity

All organisations want to increase productivity, view the Siemens case study on Improving Process Efficiency and Driving Savings 

Ensure Compliance

Financial Services and Utilities operate under strict compliance thresholds, see how Apromore Applies Process Mining to Help Diverse Vertical Industries Manage Compliance.

Challenges

In Process Manifesto¹ , the authors present 11 key challenges (thorough!), 2 of which are represented below.

Gathering & Transforming the Data

Data preparation can take up to 80% of the time spent on an ML (Machine Learning) project² .

It's no different with process mining, the bulk of the time will be spent gathering, transforming and loading the data so the algorithms can understand trends, validate workflows and suggest improvements.

Time aside, there's additional expense to acquire these resources and skills if they're not available in-house.

Benchmarking

In the paper (Process Mining Manifesto), the authors looked at bench marking from a technology viewpoint. That is, process mining has multiple techniques, however, there's a lack of bench marking to evaluate performance.

Therefore, process mining should be tied to an objective e.g.:

1. To optimise order-to-cash cycle time by analysing delays in order processing and invoicing

2. To reduce operational costs by identifying redundant or inefficient steps in core business processes

3. To enhance customer satisfaction by pinpointing and resolving service delivery bottlenecks 

Organisations can then evaluate process mining and its implementation costs against realised savings, productivity gains.

General Use Cases

For a general understanding of use cases and applications, check out the article from AI Multiple Research covering 44 use cases in 8 sectors.

Conclusion

Manual process mapping still has its place, especially for simpler tasks. But when you're dealing with complex operations and have the right tools and know-how, process mining should be a considered option.

As an analyst, it's crucial to understand how businesses actually work, not just how we imagine they do. Process mining cuts through the guesswork, showing you the real picture!

Thinking about getting started? Celonis's guide (have a look in the 'Further Reading' section) lays out three ways to begin.

You could start small with a Discovery Process Mining Project, step it up with an Enhancement Approach as you get more comfortable, or go all in with an Execution Management System for a complete overhaul.

It all depends on organisational need and resources.

Process mining isn't just the latest fad - it's a solid technology that can transform how you analyse and improve your business.

Further Reading

Additional Learning Resources

Keen! Here’s more resources:

1. Free Online Course: "Process Mining: Data science in Action" on Coursera

 Offered by Eindhoven University of Technology

2. Professional Community: IEEE Task Force on Process Mining

 A global community of academics and practitioners in the field

3. Top 11 Free Trial & Open Source Process Mining Tools

 List of providers, tool descriptions and whether they’re free trial or open source

1  Process Mining Manifesto Conference Paper

2  What is Data Preparation?