Accelerating Process Discovery with Arkturus

Part 1 of 4 – Learn how Process Mining delivers “Discovery” faster and with more accuracy, than traditional methods.


Process Mining is a new technology that transforms how business process improvement is done – from initial discovery, through to measurement, root cause analysis, and ongoing monitoring and management.

This blog focuses on the ‘Discovery’ stage and how fact-based process mining gives you a much stronger foundation, to build your process improvement cycle on.

The main objective of the Discovery phase is to get an understanding of the current state – both the steps and work undertaken in the process, and the problems that need to be addressed. 

Process mining helps you visualise your process quickly and accurately, with a complete picture of the current state, and a way to identify and quantify issues faster and with less effort than traditional methods.

Traditional Method of Discovery

Anyone who has performed process discovery will know what follows all too well. 

Firstly, to determine the current state, you try to find some existing documentation or process charts.  If you do find something, it is often out of date, so inevitably you end up starting from scratch anyway.

Typically, you arrange a workshop with stakeholders and subject matter experts of the process, then stand in front of a blank whiteboard and say, “tell me what you do”.

After several workshops and one-on-ones with key individuals, you have a documented process (probably including some Visio charts).    Unfortunately, this process documentation will be out of date as soon as any step in the process is changed… but this is the subject of a future blog.   

The real issue with Process Discovery

Here’s the real issue.  The new process documentation often includes assumptions based on what users think happens in the process. In large organisations with complex processes that span multiple teams, systems and/or locations – no single person or team know the entire process.    

Gartner is fully aware of this problem, stating:  

Traditional discovery and modeling of your operations and related processes can be a costly and time-consuming process that is vulnerable to human interpretation, lack of business knowledge and lack of objective validation techniques …

Invest in process mining capabilities to provide visibility and understanding of the actual performance of business operations and processes before starting any automation initiative, whether at a task, workplace or process level.

Market Guide for Process Mining, Gartner, April 2018

At these discovery meetings, users will also tell you about their pain-points. You have no choice but to rely on their assessment of the impact and priority of these, because at this stage you have no other information to go on. 

But what a single user feels are major problems for them, may not necessarily be major problems in the wider context of the process.  So you add the problems to an issues register and initially prioritise them based on this anecdotal evidence, which is often inaccurate.

As a result, the most important stuff is often not worked on first.  

The Discovery phase with Process Mining

Process mining changes the discovery activity completely.  It starts with generating process visualisations that show how the process is actually operating by using the event data from your systems.  This means you are dealing with facts that can’t be argued with.    

You still need to talk to users and ask them what they do and what their problems are.  But you aren’t completely reliant on their opinions or perceptions – you have much more than that at your fingertips to enable fact-based exploration.    

Rather than starting with a blank whiteboard, you can share an interactive process chart and say: “the data shows the process is currently operating like this – let’s validate it and then identify the main issues we need to address”. This completely changes the dynamics of workshops and one-on-one user meetings.    

Furthermore, it is not just a simple process chart.  In addition to the ‘happy path’, the Arkturus process visualisations show all the exception paths, bottlenecks, and re-work loops. And it is fully interactive, giving you the ability to filter by key attributes such as say ‘Location’ or ‘Product’, or by specific events or exception paths. You can easily look at example cases to better understand a problem.  

Demo data

Often while reviewing a process interactively with users, problems are identified that users either didn’t even know existed or didn’t realise the impact was so large.

On the other hand, if users mention a major problem they have, you can bring that part of the process up on the screen and find out, with them, how big a problem it really is. 

Often, what might be considered major is actually not: “Here’s the exception and it affects only 0.5% of cases, so we can assign this one a low priority”. It is hard to argue with facts. 

Discovery is faster and more accurate with Process Mining

Arkturus process mining accelerates the Discovery phase and produces much more accurate results using fact-based exploration.  

By using process visualisations based on your system data you can see how the process is actually operating – not just the ‘happy path’ but all exceptions, rework and bottlenecks. You get a more accurate and detailed view of the process in a much shorter time.  

This will further enable you to have those fact-based conversations (rather than based-on user perception) earlier in the process, and create a much more accurate issues list, usually including problems that normally would not be identified during Discovery. All this will allow you to prioritise issues faster and more accurately, and ultimately start working on the most important stuff first.  

Our next blog in this series explores how Process Mining transforms the Measurement activity.  If you would like to get future blogs delivered directly to your inbox please subscribe now.