Identifying skills that will be needed in the future, finding candidates who are willing to change jobs, or predicting illness and layoffs: the expectations placed on the analysis of people are high. What lies behind these terms? What applications are possible – and in compliance with the law? A summary.
With new technologies, data are becoming increasingly important. Its collection, transmission and analysis have advanced considerably in recent years. More and more human resource data are available and far-reaching analyses are becoming more conceivable and feasible. These new applications are discussed under the term People Analytics. But what exactly are the techniques and use cases involved? How can people analysis contribute to evidence-based personnel management? And what legal issues need to be considered?
Where does People Analytics start?
People analytics is the analysis of large amounts of data in order to be able to make information-based personnel decisions. Although people analytics seems to have become established as a common term, there are related terms such as Big Data HR analytics, HR intelligence and workforce analysis, some of which are aimed at the same methods. Furthermore, such a definition does not allow us to clearly say where HR analytics begins and where classic HR monitoring ends.
To refine the definition, the German authors Petry and Jäger therefore specify three aspects of people analysis: It is the analysis of increasing amounts of employee-related data (Big Data) through increasingly powerful data analysis options (Advanced Analytics) with the aim of securing knowledge and information and thus contributing to evidence-based personnel management.
Big Data and Advanced Analysis
In the course of digitization, the volume of data generally available is increasing, the data available is becoming more heterogeneous and the speed of data transmission and processing is increasing. In summary, these three developments are addressed by the term “Big Data” as described by Petry and Jäger.
Since the digitization of personnel data in companies and the interconnection of data on the World Wide Web also makes large amounts of data relevant to HR management available, this basic requirement for people analysis is increasingly being met.
In addition, according to Petry and Jäger, procedures for analyzing such large amounts of data have also been expanded: Through multivariate analyses, simulation procedures and data mining procedures for pattern recognition and other methods, statistical correlations can be established ever more effectively. Thus, “Big Data” becomes “Smart Data“.
Individual analysis methods have points of contact with the field of artificial intelligence (AI): For example, deep learning approaches can also be used to recognize patterns in large amounts of data. However, the use of such analysis methods does not necessarily make people analyze “intelligently”, as their use as analysis methods does not yet generate automated problem-solving competence. However, this is normally expected from an AI solution.
Since relevant data for personnel are often personal data, data protection aspects must always be taken into account: After all, available data should not be used unconditionally for all purposes.
Four levels of people analysis
In order to obtain useful information for management with the described analysis methods, analyses at four levels are required, write Petry and Jäger:
- Descriptive methods provide descriptive information: What is it and what was it?
- Explanatory methods provide information on correlations: Why is it so?
- Predictive methods provide information about probable developments: What will it be like?
- Predictive methods provide information about possible courses of action: What should/could be done?
In addition, when evaluating the results, it should be noted that not all statistical connections, nor all patterns in the data also represent a causal connection.
Evidence-based approach to analyzing people
In order to carry out such analyses on a test basis, Petry and Jäger recommend that analyses be based on logical relationships and assumptions, that appropriate measurement variables be defined for this purpose and that the process for applying and implementing the results be specified. This interaction of logic, analysis, measurement variables and process is called the LAMP approach. In the case of correlations that cannot be based on a logical hypothesis, the causality of the correlation must be checked later, for example by field experiments with a control group.
Use cases for the analysis of people
People analytics can be used in many areas of personnel management. In principle, applications are conceivable throughout the entire HR value chain.
In personnel planning, analyses could anticipate how personnel requirements will develop in the coming years and what skills will be demanded in the future. On the other hand, in recruitment, online data could show how and where potential candidates can be reached.
Analyses could also show what further training needs exist in the company and which employees are best suited for the corresponding personnel development measures or which employees are most likely to be considered for a management career.
Experts see future fields of application in the further linking of HR data with external data or financial and sales data to provide comprehensive answers to corporate strategy questions.
Legal aspects: Data protection and works council
Therefore, before using the human analysis, it must be checked whether the corresponding data analysis is also compatible with the legal regulations in force. Particular attention must be paid to data protection, but other areas of law, such as the company’s right of co-determination, are often affected as well.
From the point of view of data protection, when personal data are used, it must always be checked whether there is a legal basis for the use and processing of the data in the course of the person analysis. Therefore, unless there is another legal basis, the voluntary consent of the person concerned must be obtained for the processing of the data. In practice, this voluntary consent often causes two problems:
Voluntary consent for data analysis
On the one hand, consent must be truly voluntary. However, in the professional context, this voluntariness can often be questioned, for example when applicants fear that they will otherwise be worse off, or when managers have clear expectations or even instructions regarding consent. Furthermore, the voluntary nature of consent means that it can be revoked at any time. The data in question must be deleted immediately, which can have serious consequences for all analyses based on them.
The second problem concerns the necessary limitation of the purpose of such consent. Consent is only valid under data protection law if the purpose of data collection and processing is clearly established. Processing of data beyond this purpose is not possible in the first instance, so it must be examined how the purpose is formulated and whether the specific analyses actually serve this purpose. Therefore, open analyses are not possible, at least not if the purpose of the analysis is only apparent from its outcome and therefore cannot be stipulated in advance in the authorization.
Clarifying the right to co-determination
In the case of non-personal data, such as aggregate employee data, the legal requirements are less strict, but anonymization procedures should ensure that the data is not actually traceable to individuals.
In addition to these data protection considerations, when applying the analysis of individuals it must be checked whether the works council has a duty of codetermination. Experts therefore recommend that the works council and a data protection expert be involved in a forthcoming personal analysis project as soon as possible.
It therefore makes sense to consider together which applications of people analytics can bring the company and its employees decisive advantages in the digital transformation and, in addition, to professionally clarify all legal obstacles in individual cases.