Process mining reveals the real state of your processes

Process mining reveals the real state of your processes

Process improvements empower companies to reduce costs, improve customer and employee satisfaction, and operate more proactively.


Process mining aims to visualize and analyze real business processes. Process mining allows us to compare what is happening in reality with what has been assumed to happen in theory. In practice, business processes are more complex and less structured than the documented and expected “ideal” processes. The Purchase-to-Pay process, for instance, can be described as a step-by-step sequential process, whereas the reality – as shown by factual and historical data –  is often more complex. 

Exceptions and deviations

There may, for example, be various events occurring in the chosen process simultaneously, or in parallel, or in the wrong order. Moreover, there may be exceptions to the presumed or planned ideal process flow. Such exceptions are not only conveniently ignored by the best-case process scenarios, but are also hard to detect and understand, even though they might have a big impact. For example, the transactional data mined from an IT incident/issue tracking system may tell us that, whereas 90 percent of the transactions follow the ideal path, the deviations that constitute the remaining 10 percent cause big delays of perhaps 250 days on average. Something is wrong in these cases, and this is exactly what we are looking for. We want to find out when and where the defects occur. Exceptions and deviations often require special attention and extensive analysis, in order to find out how to correct the process so that it runs in a right and efficient way. Process mining enables us to get to the bottom of what is going wrong.

Furthermore, companies may devote a lot of their time to repeated or unnecessary work activities that were impossible to predict when planning or documenting the business workflows. Such undesired work not only affects business planning, but also harms client service, as clients might not receive the service in the time promised. In addition, company employees often see only a part of the whole process, as they fulfil their own role. They may not be able to see what happens before or after their part of the work is done, and what is the big picture. This can create bottlenecks in terms of business continuity and transferability, and the problems stemming from the whole process chain – such as insufficiency – might be missed. 

Process models are just like digital maps

As a result of process mining we are able to convert companies’ data sets, and even logs, into mathematical representations – which can then be used to create process models. This phase is called process discovery. Typically, we aim to describe “as-is” behavior, i.e. the current situation with respect to the system’s processes. In principle, we can gain an overall view of the decision points in the processes, and can investigate and understand the deviations and problems. Once we have a representative process model, we can investigate various aspects of the processes, draw conclusions and make recommendations. Process models can be seen as digital maps – once we have an interactive online map, we can zoom in on specific parts (e.g., over a country or a city), switch between various views or analyses (e.g., satellite view, bicycle roads or traffic) and observe the reality. No single map can show all aspects of the process, and therefore we need to conduct multiple analyses. After these analyses are completed, we can identify areas that need improvement (e.g., congested roads), the times when they occur (e.g., on the weekends) and why they occur (e.g., poorly designed roads).

Process mining produces improvements

As we visualize and analyze the whole process flow, the process becomes more transparent and we can capture concurrencies, exceptions and unnecessary work. Then we can begin to plan for process improvements that move towards the desired state, based on factual data. For instance, there are many ways to carry out an invoice process. Through data mining we can, for example, see why some invoices related to purchase orders are always difficult to match. It is also possible to see, at which step of the process, the Accounts Payable Clerk is using a lot of time, and what solution could be found to prevent this happening in the future.

Furthermore, we can:

  1. identify those parts of the processes that could be automated
  2. understand and classify ERP system error messages, and
  3. analyze the order of transactions.

Process improvements empower companies to reduce costs, improve customer and employee satisfaction, and operate more proactively. 

More information

Sezin Yaman

Data Scientist, KPMG Lighthouse

+358 20 767 2181


Sanna Leskinen

Finance Strategy & Transformation, KPMG

+358 20 767 2233

© 2024 KPMG Oy Ab, a Finnish limited liability company and a member firm of the KPMG global organization of independent member firms affiliated with KPMG International Limited, a private English company limited by guarantee. All rights reserved

For more detail about the structure of the KPMG global organization please visit

Ota yhteyttä