Now you know what Process Mining is about, let us see how you may generate and deliver even more value through this promising (data science – based) technology. This first part of the blog mainly focuses on value cases made possible thanks to the “process discovery” discipline, also including some cases for process improvement.
Explore your real processes fast and accurately
Imagine that – for any reason listed here below – you want to know how your current business processes really look like; rather than “as assumed”. As event logs originate from activities and transactions that really occurred, the reconstruction of process diagrams through Process Mining reflect even better the real processes. Moreover, once you have access to the relevant event logs, process modelling is faster and more efficient.
Process Mining software enables to produce different types of process models from event logs, incl. BPMN models, EPC diagrams, Petri Nets, etc. Some applications (like ProM) even allows you to export BPMN diagrams as XML files, which might be used for other purposes in other software applications. The image at the right illustrates a diagram resulting from the “Fuzzy miner” algorithm of the Process Mining software Disco (Fluxicon).
Sound automation investments
Too often, large investments in business wide projects – ERP, CRM, SCM,… implementations and alike – are made by companies not knowing (thoroughly enough) their ‘as-is’ business processes. This does not only lead to a waste of the investment money, but even worse: it negatively impacts future company activities, with lower efficiency and less customer satisfaction as a result. Like explained in the blog of 16 April 2015 (particularly, § “Reflect before you automate”), automating non-optimal business processes will not help to optimise your business; on the contrary.
It is no news that a profound understanding of an organisation’s business processes is a critical success factor for ERP – and the like – implementation projects.
Can you imagine how much value a company may save by knowing in depth its real ‘as-is’ processes before deciding to invest in expensive and burdensome automation projects?
Lower software maintenance costs
A similar observation is that (future) users of enterprise wide software often demand expensive customisations for activities that are actually rather exceptional, thus for rarely used process variants. Customising such software for very specific process variants – which are executed once in a while – is not only very costly during the implementation. This is a really expensive – above all recurring – burden at each software migration, that may need to happen every few years if you want to keep the support from the software provider. This blog confirms this experience of costly maintenance due to software customisation.
Can you estimate how much value an organisation may save on the long-term by avoiding such value decreasing customisations?
Above graph illustrates how Disco reflects the number of process instances (= cases) for the respective process variants. It reveals that 5 of the 116(!) process variants cumulatively represent 80% of the cases (process instances). A classic example of the Pareto principle…
Below image illustrates the similar finding, though more explicitly how many cases there are for each process variant. It is actually quite the same histogram as here above, though rotated by 90° clockwise.
You may guess for which part of the process variants it is sensible to customise software. An experienced Business Process Expert will investigate why all these process variants differ from variant 1, which represents about the half (i.e. 713) of all cases. And this expert should also assess whether the cost of maintaining other process variants is economically justified.
Smarter integration for Mergers & Acquisitions
Imagine that your company acquires – or merges with – another one. The rationale behind Mergers & Acquisitions most often being economies of scale, this means that both organisations should converge to work the same way. Hence, the aim should be to align their business processes as much as possible.
Wouldn’t it be valuable and easier to know how both organisations really work, i.e. knowing their real ‘as-is’ business processes, so to align their activities to common best-practices?
Decrease variability and standardise processes
Discovering process variants is not only useful for enterprise wide software – or other organisational – projects. It also facilitates continuous business (process) improvement activities. Seasoned Business Process professionals know how variability negatively impacts process performance. The 2 cases here below illustrate how Process Mining facilitates the identification of variability.
Variability in ‘flow’ (process throughput time)
Lean professionals will recognize the importance of “flow” – sometimes called “Lean Flow”, within a process. This actually means that the process throughput time should be as short as possible. Below example illustrates how process mining tool Disco reflects the spread in process throughput times thanks to its functionality “case duration”. It shows that 31% of the cases are run through, within a period of 2 days and 19 hours, while the longest lasting case takes more than 278 days.
This is very valuable information for the process professional who has a nice opportunity to investigate why there is such a huge spread in throughput times.
As we will see further, these process mining tools also allow to identify where (for which activity) precisely in the process a specific process instance spent lots of time. This obviously helps to figure out the underlying causes for these variations.
Variability in process input or process triggering
Knowing the delays within a process is one thing. However, it is also useful to know the variability of the process input – or similarly, the variations of what triggers the process. The image here at the left is a so called “dotted chart” obtained in ProM, illustrating the arrival of requests for environmental permits at a municipal administration. As one can see, the rate of arrivals – represented by the blue dots forming a diagonal – is quite regular over time. But for less regular rates, it would be valuable to identify the variability patterns, so to understand their origin, and possibly how to manage the input variability.
Don’t miss the second part of this blog on the value cases of Process Mining, which will further describe cases on process improvement, and also will explain many reasons why you may use process mining for conformance checking as well.
To make sure you don’t miss the second part of this blog – or any further blogs -, you may still subscribe to 2-weekly “process optimisation tips”, by registering via the “Stay informed” box at the right side of this page.
P.S.: Please don’t hesitate to share this information with your Facebook friends and fans, LinkedIn contacts, Twitter followers and Google+ circles, through the share buttons below. Thank you!