The promise of process mining is using large volumes of data to analyse and optimise processes. Are these tools a key to value creation?
As CFOs are keen to optimise business processes in order to improve performance and to control risk, the explosion of available data volumes may be an opportunity to seize.
The statement is commonplace by now: data are the new gold, the engine of the fourth industrial revolution. This is true, but data are valuable only if they can be turned into information, valued as knowledge and used to guide decisions. For a company, the data most easily accessible and (supposedly) best controlled are those collected and stored in its own information systems. Such data are often difficult to use, however, as evidenced by the efforts required to ensure regular reporting of performance indicators. The reporting process is tedious, and the information is sometimes out of date before it is even published!
Many companies have turned to process mining to try and capitalise on internal data, precisely in order to understand the company’s processes better. The variety of use cases identified shows that this technology is part and parcel of the digital transformation of companies to put data at the service of process optimisation and business performance.
Process mining: a load to be followed to unearth the gold hidden in the data of our systems?
The company’s value chain is governed by processes: order booking, shipping, returns, invoicing, production, innovation…
The departments in charge of their execution have implemented a number of indicators that enable them to measure the performance and cycle times of the missions they are entrusted with. Today, many steps are digitised and recorded in the various information systems, and each action taken leaves traces therein. It is these digital traces or “logs” that process mining (process science + data mining) proposes to capitalise on.
Process mining, i.e. “process exploration,” brings together various data and process analysis techniques.
Since the turn of the millennium, we have witnessed the parallel development of process analysis on the one hand, with Business Process Management (BPM) techniques, and data analysis on the other, with the emergence of Data Science, which uses statistical models to facilitate the analysis of large volumes of digital data. Unfortunately, these two fields did not speak to each other. According to Professor Will Van der Aalst, one of the pioneers in the field, the contribution of process mining is to create the “missing link” between the two disciplines. Process mining uses specialised algorithms to identify trends, patterns or correlations between data so as to gain a better understanding of processes, especially their efficiency.
What are the fields of application of process mining?
Process mining can be used in many ways and can be grouped into three main areas:
1 Understanding and modelling your processes using data:
In a conventional approach, as promoted by Lean Six Sigma methods, we start by compiling data to define and measure the problem to be improved. Data constitute the necessary entry point for any continuous improvement process.
In today’s context, bringing together the best experts face to face in workshops has become a challenge, while gathering reliable data manually to identify best practices remains a complex undertaking: the data are difficult to compare, the systems are different, the practices vary, and yet, it is all about the same process!
A modelling exercise requires a significant presentation and validation effort. Process mining algorithms can automate the process modelling stage from the data in our systems.
Process mining is based on an exhaustive approach to data. By connecting directly to the systems to be analysed, all the logs are used to model the process that is actually carried out. This is a considerable advancement! Not only does it speed up this phase, but also the results form an indisputable basis for identifying best practices and defining a target because real data are used. The use of process mining makes it possible to link the data to the process. The necessary vision of how to structure harmonisation or optimisation approach can be obtained by gathering the different execution indicators. Once equipped, the process becomes clearer and the decision is guided by the data.
2 Data transparency makes it possible to identify areas for performance improvement
Process mining is a facilitator that simplifies data production and analysis while identifying areas for improvement in the company’s key processes.
These tools produce performance indicators automatically. Substantial time is saved at all stages, as data extraction, reprocessing, consolidation and presentation are automated. Process analysis becomes dynamic, visual and intuitive, capable of selecting all or part of the data, comparing the performance of different perimeters, identifying trends and performing causal analysis.
The same tool can be shared between operators, analysts and decision makers, depending on the (operational or strategic) level in the company. Activity reviews become more efficient, as questions raised can be answered in real time thanks to data visualisation. Process mining draws on data from the processes to highlight areas for improvement and to reveal sources of productivity. Bottlenecks are clearly visible and represent opportunities to be seized by the teams in charge of automation projects.
3 Data at the service of internal control:
As we have seen, the strength of process mining lies in enabling rapid access to all the events in a process. This is an important change for internal control approaches. It is no longer necessary to limit controls to a representative sample of data. Certain controls can be generalised to the entire database. By modelling the target process, process mining tools identify deviations and can launch alerts whenever a significant such deviation is identified, thereby reducing the associated risk.
Process mining enables the organisation to learn and progress by avoiding repeating past mistakes and by assessing the impact of a change by replaying past data.
CFOs can therefore avail themselves of the power of process mining algorithms to advance three major priorities:
- Performance management;
- Process optimisation;
- Risk management.
Which process to start with and which skills to mobilise in order to launch a process mining approach?
The areas of application and associated use are legion:
- Finance: customer or supplier payment deadlines;
- Supply chain: inventory supervision;
- Customer service: improving the cycle from order to delivery.
The performance indicators can be adapted to the targeted process (customer journey, supplier relations, product tracking, etc.). The link with the challenges and priorities obviously has to be made in order to identify the key process(es) to be analysed. Sponsorship by management and the strong involvement of those responsible for the process are consequently necessary to ensure that the analyses will be truly beneficial.
The problem is reversed when the expected benefits are considered: Which indicators diverge without me grasping the root cause thereof? What are the benefits for my automation projects? Which stage requires better risk control? If the answers to these questions converge on a particular issue, it’s probably a good starting point to put your company’s data to work!
An agile methodology is frequently observed during the implementation phase. Limiting the scope enables us to target the events to be analysed better and to scale up iteratively by replicating the approach on new scopes. But a clear scope is not enough to launch a process mining approach. In projects dealing with data, the complexity will increase rapidly if the data are not structured or do not comply with clearly defined management rules. Technical problems can also be encountered, for example if the system in question does not keep records of events at the right level of detail. Many management indicators simply do not require such techniques, in fact! It is the dynamic view of data and processes that makes process mining particularly relevant.
But the key to success lies in the teams mobilised first and foremost. If a process mining project does not require a team of data scientists, knowledge of the process and the associated systems is essential to ensure the success of the project and to create the link between events and data.
Conclusion
Faced with an explosion in the volume of data generated by our systems, the data strategy can now include an operational performance component based on process mining. The increased power and growing precision of algorithms offer companies an opportunity to leverage their own data and changes the way they approach the optimisation and control of business processes. Process mining makes it possible to give meaning to data and to create value around the company’s processes. As with any project aiming to process data, it is important to define a clear objective and to follow a suitable methodology so as to avoid getting lost. But the first thing to do is build a team around the people who master the process and the associated systems. They will be the real alchemists, the only ones capable of turning system data into gold or, at least, into cash!
Guillaume Siccat is a key contributor to the RPA/Intelligent Automation committee of DFCG chaired by Armand Angeli. First published in Finance & Gestion, the Magazine of DFCG, the French CFO institute
