Business is becoming increasingly dynamic and complex. Technologies and working methods are currently available to support the decision-making process and deal with the new challenges. In particular, process and data mining is a highly effective tool for CFOs.
What is Data Mining?
Process Data Mining is a tool that supports the analysis of company data held in ERP, CRM, PPM, and other such systems. Based on the chronology of events that exists in the systems, it virtually reconstructs business processes, thus allowing the AS-IS processes (and the respective variations) to be analysed and the potential areas for improvement and, as a last resort, automation, to be identified.
Data Mining makes extensive use of algorithms, Artificial Intelligence (Machine Learning), Data Warehousing (an organised collection of company data) and Data Lake (a mass repository of various kinds of internal or external data) technologies.
What is the purpose of Data Mining?
To provide timely responses, recommendations, and forecasts, and essentially to facilitate the decision-making process. Observing, guiding, deciding, and acting are the steps that map any decision-making process, including that of the CFO. Selecting the best combination from a wide range of existing technologies is not a trivial matter. It is essential, however, to be clear that the decisions we make today must be based not only on an awareness of yesterday’s situation, but also on the “here and now”.
Using these tools allows us to analyse company data in different ways, the main ones being:
– Process Discovery: AS-IS business processes are analysed using the event logs in the company systems;
– Conformance Checking: processes can be compared with a reference model in order to analyse the differences;
– Process Enhancement: the objective is to make the process more efficient by highlighting the potential areas of automation. The main characteristics of Process Mining tools that facilitate the analysis of company processes are listed below;
– Transparency: allows the end-to-end process to be analysed with the option to go into the details of the individual process steps;
- Extraction of event logs from the various source systems.
- Collection and organisation of event logs
- Processing and representation of processes
- Process analysis
– Performance: process KPIs can be defined and analysed with the option to verify the completion times of the various process stages, identifying any “bottlenecks”;
– Conformity: allows the real process detected with the model process to be compared, identifying deviations and potential violations;
– Organisational analysis: allows the organisational aspects of the process to be analysed, determining, for example, the number of resources employed on the various process steps and analysing the activities that could potentially be automated.
Furthermore, different levels of decision-making automation can be pursued based on the degree of complexity of the context to which the decision relates. Looking at this in detail, we can distinguish three different cases:
– Decision Automation: observation, guidance and decision-making activities are fully automated. Execution is often semi-automated as another process may be activated immediately afterwards or the actors may be notified to perform a specific action;
– Decision Augmentation: (or Augmented Decision Support): observation and guidance are automated, while the decision is semi-automated. Then, using prescriptive algorithms, the system offers suggestions/recommendations to the user, who always makes the final decision.
– Decision Support: only observation is fully automated in terms of data collection, while the other stages of the decision-making process are the responsibility of the user. It is also important to underline the increasingly strong correlation between Data Mining and the new concept of Continuous Intelligence which, as defined by Gartner, is:
“A design pattern in which real- time analytics are integrated into business operations, processing current and historical data to prescribe actions in response to business moments and other events”.
The basic concept is that every automated decision support process, at any level, must be continuous and integrated into company processes and not delegated or “relegated” to special one-off projects.
A few figures
Companies that use Big Data in an automated and integrated way can increase productivity by 5 to 10 percent (European Parliamentary Research Service).
According to other estimates (MGI, McKinsey), if we can harness the great value of “data”, learning lessons and gaining the awareness to make better decisions, the manufacturing and commercial sectors could increase their margin by 60%. In Italy, however, it appears that only 7 percent of companies use Big Data to support decision-making in a structured, integrated, and continuous way (2014-2020 data, European Parliament – IDC).
Data Mining to support business
Before delving further, it is worth dwelling on the main barriers and risks that often accompany the introduction of Process Data Mining:
– identifying the appropriate technologies to automate processes in a way that is appropriate for the company’s organisational context can be a particularly complex task;
– determining the business processes and operations on which to focus this methodology requires time and money, and this often causes an impasse in business processes;
– without governance planning, there is a risk of derailing or slowing down the corporate digital transition strategy;
– this technology is based on an analysis of company databases, the more the processes are managed within information systems, the more effective these technologies are.
It might therefore be helpful to:
– define KPIs intended to identify costs, improve market reach (business volumes) and internal efficiency (margins), to first and foremost obtain a better (and measurable) return on investment;
– focus the Process Data Mining on assessing the efficiency and effectiveness of processes as quickly as possible without human prejudices, in order to address and critically evaluate any opportunities to implement automation technologies or carry out the necessary reengineering work;
– use Process Data Mining also to fully support process governance.
Process Mining User Guide
Standard operating procedures, policies, work instructions or best practices integrated into business applications are often “muddied” by informal activities, which lead to governance problems and compromise the expected value of these business applications. Process automation (Robotic Process Automation – RPA) and the extensive use of new technologies (IoT in primis) can provide a significant benefit in terms of effectiveness and efficiency.
How can CFOs master these emerging trends and guide their strategic decisions?
To begin with, they must:
– invest in adequate Process Data Mining skills to gain visibility and an understanding of company performance and processes before starting any automation initiative;
– assume a leading role in educating colleagues on the advantages of using Process Mining tools;
– research and explore use cases that go beyond the analysis and validation of basic processes.
From RPA to hyperautomation
It is therefore important to define what the “digital ambitions” of your company are and what objectives it wants to achieve.
To be concise, we can restrict the areas of improvement using the classic paradigm:
– Revenues: what are the key factors that influence your company’s revenues? For example, you can opt to improve the process of evaluating opportunities rather than automating activities, increasing customer involvement rather than introducing new services;
– Costs: how can costs be optimised? Efficiency can be improved by automating activities, but also by redesigning processes, reducing the cost of errors and speeding up processes;
– Risks: what are the risks arising from inefficient processes? By redesigning and automating processes, the risk of non-compliance with a regulatory process can be mitigated.
In practical terms, every company must therefore:
– identify the situations they want to optimise in order to improve the effectiveness and efficiency of a process;
– aim to transform company processes, particularly by experimenting with new ways of providing value and reprogramming and optimising existing processes;
– clearly distinguish in its roadmap the use cases from the desired results, dividing these impacts among the areas previously described, i.e. higher revenues, lower costs and risk mitigation.
Understandably, this process strongly characterised by innovation: by 2022, 65% of the organisations that have implemented robotic process automation will have introduced Artificial Intelligence, including Machine Learning and Deep Learning, and natural language processing algorithms (source: Gartner, Move Beyond RPA to Deliver Hyperautomation – Authors Saikat Ray, Cathy Tornbohm, Marc Kerremans, Derek Miers).
The new role of the CFO
The role of the CFO, as ANDAF has often pointed out, is changing very rapidly. From a specific (accounting driven) technical characterisation, founded on knowledge of the basic administrative rules (financial statements, accounting principles, taxation, etc.) it is being transformed into a Technology Driven role, which is the only way for the CFO to become a business advisor/partner capable of supporting every aspect of corporate decisions, necessarily leveraging new Information Automation technologies, and Process Data Mining in particular.
In this new context, the CFO will, for example, be able to fully support the analysis and evaluation:
– of the potential inclusion of a new technology in the product/process (based on an analysis and correlation of data regarding internal R&D projects and on information from the analysis of external competition collected by the strategic planning function); if made too early or too late, this decision can lead to an unprofitable investment or a loss of competitiveness on the market;
– of any business process (based on modern digital technologies) in terms of efficiency, identifying bottlenecks or gaps to be filled by extracting particular critical behaviours, through statistical modelling, and subsequently identifying the causes of recycling, bottlenecks or waiting times;
– the impact of a supplier on company project deliveries in terms of on-time delivery and the need for assistance and maintenance for parts or components;
– acquisition or merger operations using Process Data Mining as a crucial starting point to guide reorganisation and change operations, allowing effective and reliable programmes to be established for the organisational harmonisation and standardisation of systems.
More generally, the Finance function also plays an important part in Performance Management. In fact, among the most important tasks performed by administrative and financial officers, beginning with the CFO, are the management, control, and analysis of company performance. The ability to create added value for the company now lies in the hands of the CFO, providing tools and information to support the decision-making processes performed by managers, increasingly involved in competitive challenges that continually redefine the concept of performance.
Furthermore, when senior management changes take place, Process Data Mining techniques allow the new managers to understand the company’s operating model (objectively and in a short time). They can therefore introduce Performance Management tools to establish improvement objectives and measure the results gradually achieved. Risk management and the need to protect the company from changes in economic scenarios can also rely on Process Data Mining techniques. Just think, for example, of the role of the CFO as the Dirigente Preposto (manager in charge) pursuant to Italian Law 262/2005, and the opportunities that these tools provide to support compliance and auditing checks, revealing how a process is carried out in reality, constantly highlighting activities and operating procedures, and determining whether or not they comply with established policies and procedures. In this respect, it also appears to be one of the tools that is spreading most widely within corporate auditing structures.
Finally, Process Data Mining allows the rigidities between the various company departments to be removed, speeding up the collection and connection of data from disparate sources and, above all, supplementing the budgeting process and mere accounting indicators with non-accounting information that truly represents the company’s performance. Having top-down and bottom-up information in real time is an advantage as it provides knowledge of the entire workflow, quickly identifying potential internal and external threats. This also applies to the management of responsibilities related to compliance with industry rules and regulations. Monitoring the risks arising from non-compliance with legal provisions, especially in a system as intricate as the Italian one, can in fact be difficult without dashboards and scorecards that can be read and understood immediately.
As previously pointed out, the role of the CFO is changing and becoming that of a true business partner. By using Big Data, the CFO can provide increasingly relevant insights into all the activities that generate value for the company. The expansion of the CFO’s skills depends not only on the ability to implement these new digital tools and the relative ability to analyse increasingly complex data, but also on his or her ability to translate the information received through these “enabling technologies” into strategically significant indications.
Donato Pastore is Chairman, ANDAF ICT Technical Committee, Procedures and Systems Manager, Ferrovie dello Stato Group
Luca Lucidi is Member, ANDAF ICT Technical Committee, CFO & Investor Relations Manager CY4GATE
Previously published in Italian by the monthly magazine of ANDAF, the Italian CFO association.
The authors are speakers at the CFO Automation Experience Round tables