UNIVERSITY OF VAASA SCHOOL OF MANAGEMENT Minna Lanu BUILDING HR ANALYTICS MATURITY Case Study Master’s Thesis in International Business Master’s Programme in International Business VAASA 2020 1 TABLE OF CONTENTS Page LIST OF TABLES AND FIGURES 5 ABSTRACT 7 1. INTRODUCTION 9 1.1. Background of the study 9 1.2. Research gap 11 1.3. Research questions and objectives 12 1.4. Structure of the study 13 2. LITERATURE REVIEW 14 2.1. Evidence-based HRM 14 2.1.1. From data to big data 19 2.1.2. HR metrics and measuring 22 2.2. HR analytics 27 2.2.1. Defining HR analytics 27 2.2.2. The levels of HR analytics 31 2.2.3. The people behind HR analytics 38 2.2.4. The processes of HR analytics 42 2.2.5. Technology in HR analytics 45 2.2.6. Data in HR analytics 48 2.2.7. Governance in HR analytics 50 2.2.8. Theoretical framework of the study 52 3. METHODOLOGY 53 3.1. Research approach 53 3.2. Philosophical assumptions 54 3.3. Research strategy 56 3.4. Data collection 58 3.5. Data analysis 61 3.6. Reliability and validity of the study 62 3 4. FINDINGS AND DISCUSSION 64 4.1. Views on HR analytics in the case company 64 4.2. People and their analytical skills and competencies 68 4.3. The necessary building blocks behind HR analytics process 73 4.4. Technologies supporting HR analytics maturity 77 4.5. Data and its governance 81 4.6. The future of HR analytics in the case company 86 5. CONCLUSIONS 89 5.1. Key findings of the study 89 5.2. Managerial implications 92 5.3. Theoretical contributions 93 5.4. Limitations of the study and suggestions for future research 94 LIST OF REFERENCES 96 APPENDIX 106 5 LIST OF TABLES AND FIGURES Table 1. Examples of efficiency, effectiveness, and impact metrics 25 Table 2. Examples of human capital metrics 26 Table 3. The characteristics of the case company at the end of 2019 58 Table 4. Interview details 60 Figure 1. From a professional practice into a decision science 17 Figure 2. The HR intelligence value chain 18 Figure 3. The levels of HR analytics 33 Figure 4. Degrees of analysis and business intelligence 34 Figure 5. Maturity levels of HR analytics 36 Figure 6. The IMPACT Cycle 43 Figure 7. Theoretical framework of the study 52 7 UNIVERSITY OF VAASA School of management Author: Minna Lanu Topic of the Thesis: Building HR analytics maturity Name of the Supervisor: Vesa Suutari Degree: Master of Science in Economics and Business Administration Master’s Programme: International Business Year of Entering the University: 2013 2020 Pages:106 Year of Completing the Thesis: ABSTRACT The focus of this thesis is to study how organizations can develop their Human Resource (HR) analytics in order to better their readiness to provide valuable insights to support management decision-making. To strengthen its strategic role in the company, HR function must, instead of continuing to base their decisions on former experience and intuition, become more evidence and data-based. The HR function has to expand its offerings from simple reactive reporting to more advanced and predictive analytics. This paper is a case study conducted on a large, Finnish based, multinational company, operating in the industrial goods and services –industry. The study embraces a qualitative research approach and the objective is to identify the factors effecting and also hindering the development of HR analytics maturity. Moreover, the paper seeks to figure out the current status of HR analytics in the case organization and to provide suggestions on focal areas requiring attention and actions in order to increase the HR analytics maturity. In order to be able to produce insightful analytics from HR processes, the case organization should, first and foremost, strengthen the data culture among its HR population. Furthermore, the company should decide what it wants to accomplish with its analytical capabilities. Only after that, based on the clarified needs, the firm should start to build a HR analytics team consisting of people with different backgrounds and skills. The individuals in the team should be enthusiastic about the topic, have good intuitive skills, and be able to detect patterns in the information. At the beginning the team may conduct descriptive analytics to describe the current situation at the company. Once the team becomes more experienced and skilled it can move on to more predictive analytics to guide management’s decision-making. Additionally, alongside with the assembly of analytics team, the organization should also pay attention to its data governance. In order for analytics process to function efficiently, data governance cannot be ineffective. Thus, the case organization should establish practices to guarantee the quality of data inputs used in HR analytics. KEYWORDS: HR analytics, evidence-based HRM, HR analytics maturity, predictive HR analytics 9 1. INTRODUCTION This thesis investigates the phenomenon of HR analytics in one particular multinational company (MNC). This first part of the paper presents the background and motives of the study, as well as points out the research gap in the field by shortly presenting the main findings from the previous studies. Moreover, research questions are introduced and the structure of the thesis is described. 1.1. Background of the study Employee and workforce related insights are arguably the greatest competitive advantage for companies operating in the market where the disruption and uncertainty drive drastic changes (DiClaudio 2019). The digitalization of human resource management (HRM) has brought opportunities for human resource (HR) professionals to use the data generated by technologies to support insightful decision-making (van den Heuvel & Boundarouk 2016; Dahlbom, Siikanen, Sajasalo & Järvenpää 2019). It was proposed, already over a decade ago, that in order to enhance the quality of decisions about human capital, HR should extend its focus from the traditional service- orientation to being a decision science. With such a paradigm shift that functions like finance and marketing have already gone through, the HR function could really experience what it is like to be strategic. (Falletta 2014; van den Heuvel & Boundarouk 2016.) Due to the fact that human behavior is considerably more complex and much less predictable compared to other tangible resources, it has always been difficult to optimize the human capital allocations (Walsh, Sturman & Longstreet 2010; King 2016). “The use of HR analytics has noticed a recent rise in popularity in response to this challenge.” Through the usage of data and metrics to plan, assess, and implement new management policies, it is possible for the HR function to wean itself from using only experience, intuition, and guesswork to guide HR strategy. (King 2016.) In other words, the HR function becomes a strategic partner by providing insights gained 10 through analytics. These insights can be used to develop competitive strategies. (Bassi 2011.) However, the use and development of analytics in HRM has not kept pace with the organizational need. That arguably means that the strategic partnership of HR cannot be considered as self-evident. (Lawler, Levenson & Boudreau 2004.) HR analytics should go through a transformation from the basic measurement of internal HR metrics to concentre more on analyzing the critical human capital issues. It has been studied that organizations that sacrifice the needed time and effort to manage their staff as the valuable asset outrun their peer companies and achieve a competitive advantage. Even so, only a fraction of the top organizations have managed to move to Analytics 3.0, leaving most of the companies still executing analytics 1.0. (Vargas, Yurova, Ruppel, Tworoger & Greenwood 2018.) Many HR minded people forecast a bright future for HR analytics (van den Heuvel & Boundarouk 2016). “The growth of analytical and evidence-based decision making (and the technical tools accompanying it) has great potential for improving organizational effectiveness and efficiency” (McIver, Lengnick-Hall & Lengnick-Hall 2018). Yet, like already stated, the reality today is that organizations are struggling. Companies are having difficulties to implement HR analytics into their organizational reality. It is even suggested that HR analytics represents one of the most central capability gaps in today’s HR practice. Even the large multinational companies (MNCs) do not have clear future visions for HR analytics within their organizations. (van den Heuvel & Boundarouk 2016.) Companies lack the understanding on how to successfully use HR analytics to influence organizational outcomes. Many firms build HR analytics teams in hopes of gaining strategic insights from their people data. However, compared to other functions, the execution of HR analytics tends to fail and fall short of expectation. (McIver et al. 2018.) The challenge with HR analytics today is the lack of analyses that guide business leaders to address the top priority issues hindering the successful strategy execution (Levenson 2018). One reason behind this is most probably the lack of information on 11 how to translate ideas into practice, as most of the literature around HR analytics is more promotional than descriptive (King 2016). HR functions are nowadays spending considerable amounts of time and effort to produce descriptive reports after another. Descriptive reports are undoubtedly very beneficial for businesses to ensure that managers are aware of what is going on in the organization. However, descriptive reports are limited to present only snapshots of what is happening in the organization at a given time. In other words, descriptive reports lack the capability to help companies to make future predictions and they do not assist in understanding and accounting for why things are occurring in the organization. (Edwards & Edwards 2019.) If an organization wants to understand outcomes and predict possibilities and events, more advanced analytics must be applied (Fitz-enz 2010). 1.2. Research gap The interest toward new analytical opportunities, enabled by digitalization, has increased considerably in recent years (Dahlbom et al. 2019). There has also been a huge amount of blog posts, white papers, consulting and press reports on HR analytics. Nevertheless, it seems that the topic has gained limited attention among management researchers (Marler & Boudreau 2017). It is stated that the published evidence supporting the supposed value of HR analytics is quite scarce (Rasmussen & Ulrich 2015; van den Heuvel & Bondarouk 2016) and based more on beliefs than evidence. They are also “often published by consultants with a commercial interest in the HR analytics market”. (Rasmussen & Ulrich 2015.) Due to the lacking information on how to translate ideas into practice (King 2016), HR professionals have continued to work in a non-data-driven manner. (Dahlbom et al. 2019.) As a matter of fact, only a minority of organizations report effectively using HR analytics (Deloitte 2019) and even the big multinational corporations (MNCs) that have spent enormous amounts of money and resources in HR analytics and have achieved results in analytics in other areas of 12 business admit that their HR analytics programs are stuck in reporting historical information only (Angrave et al. 2016). Moreover, the research on HR analytics in the Nordic countries, yet alone in Finland, is arguably scarce, almost nonexistent. According to the few studies conducted on organizations operating in the Nordic countries, the topic of HR analytics is considered as very important (Deloitte 2017b; Nordic HR Study 2017), but only a small percentage of organizations report being satisfied with their HR department’s current performance in it. This indicates, according to the Nordic HR Study (2017), that the need to better the performance in HR analytics is recognized, but not yet achieved among Nordic organizations. In fact, organizations are struggling to fully exploit HR analytics and are suspended in the descriptive stages of analytics. (Nordic HR Study 2017.) 1.3. Research questions and objectives In order to better understand the ways in which data and analytics are exploited by the HR function today, and to identify the main challenges hindering it use and maturing, a case study was conducted to answer the following research questions. The main research question in this study is: • What are the key-factors in building HR analytics maturity? The main research question is supported by the following sub-questions, guiding both the theoretical and empirical parts of the thesis: • What is the current status of HR analytics in the case company? • What are the factors hindering the maturing of HR analytics in the case company? 13 1.4. Structure of the study This paper consists of five main chapters. The first part is introduction and it presents the topic of the thesis, its background, and the existing research gap in the field. Moreover, the aim of the study as well as the research questions and objectives are introduced. Thereafter, the paper moves on to literature review in which the prior research on the topic is presented by dividing it into smaller sections. The theory part proceeds in a “bottom-up” manner first discussing the context in which HR analytics occurs and then going deep into the topic itself and the factors affecting it. The final outcome of the literature review is the theoretical framework used in this study. The third part, in turn, introduces the methodological choices for the thesis. This part considers the research approach, philosophical assumptions, research strategy, data collection and analysis, as well as the reliability and validity of the study. The case company is also described in the third part of the paper. Thereafter, in the fourth part, the empirical findings made from the gathered data are presented and discussed. Finally, in the last part of the paper the main findings are summarized and study’s theoretical and managerial contributions are presented. The last part of the thesis also considers the limitations of the paper and gives suggestions for future studies. 14 2. LITERATURE REVIEW In order to be able to answer the research questions and objectives of the study, it is important to have good theoretical base to start with. In this part of the paper the topic at hand is examined from a theoretical viewpoint, meaning the most central observations from the already existing literature are presented. The chapter starts by first discussing the more comprehensive concept of evidence-based human resource management (HRM) and thereafter moves on to view the topic of HR analytics on a more detailed level. 2.1. Evidence-based HRM The practice of HRM has gone through significant changes over the past decades. HRM is transforming from a lower level, administrative and maintenance oriented function into a strategic business partner that is seen as a core organizational operator. (Beatty, Huselid & Schneier 2003; Ulrich & Dulebohn 2015.) One of the key factors in this transformation is the change in decision-making habits – HRM has/should become more evidence-based. (Ulrich & Dulebohn 2015.) Evidence-based practice means conscientious, thorough, and careful use of evidence from multiple sources in decision- making (Barends, Rousseau & Briner 2014). Therefore, “without rigorously tracking HR investments and outcomes, HR decisions and priorities remain whims not science.” (Ulrich & Dulebohn 2015.) According to several scholars, data processing, analyzing, and measuring are the main enablers of the strategic HRM decision-making. Business actors who succeed to manage their people resources in a more strategic manner are considerably more likely to gain greater financial returns. (e.g. Boudreau & Ramstad 2007; Walsh, Sturman & Longstreet 2010; Levenson 2018.) However, according to Boudreau & Ramstad (2007), organizations and their leaders do not always recognize the importance and potential of HR function and therefore fail to staff the department with people who posses the right know-how and talent. Instead, they still tend to rely on the traditional definition of HR’s 15 mission, which is “to be a respected business partner, helping the company achieve its goals by providing outstanding services to help manage the company’s most important asset, its people”. (Boudreau & Ramstad 2007: 8-9.) In order to break out from this outdated paradigm, HR should pay more attention to the decisions it supports instead of concentrating purely on the services it provides. Derived from the aforementioned statement, the new modernized HR mission paradigm for organizations to pursue is “to increase the success of the organization by improving decisions that depend on or impact people”. (Boudreau & Ramstad 2007: 9.) Ulrich & Dulebohn (2015) are on common ground as they state that it is time for HR to shift from the inside/outside approach to the outside/inside approach. This means that instead of just serving employees or redesigning and enhancing HR practices, focus should be on making sure that the services HR offers inside the company are in line with the expectations coming from the outside. (Ulrich & Dulebohn 2015.) However, the nature of HR decision-making is challenging which makes also its strategic implementation considerably difficult. HR assets, i.e. people, differ greatly from most of the other resources a company has. That is due to their intangibility. (Walsh et al. 2010.) Lev (2001) defines an intangible asset as “a claim to future benefits that does not have a physical or financial embodiment” (Pease, Byerly & Fitz-enz (2013). Examples of intangibles in the field of HR are, among others, leadership, engagement, culture, commitment, and employer brand (Fitz-Enz & Mattox 2014: 18). It is really hard to forecast human performance and therefore it is also challenging to make strong claims about future benefits a particular employee investment has. For example, if an organization invests in a new training program, it is considerably difficult to indicate its effects on employee performance, not to mention the financial value it brings to the organization in the long run. According to Walsh et al. (2010), it is like looking through a blurry lens when trying to analyze employee’s financial value for the company. (Walsh et al. 2010.) On the other hand, costs associated with workforce are more or less clear. That combined with the fact that HR often lacks the evidence for human capital outcomes 16 puts its initiatives in real jeopardy, as it may be treated as an expense rather than an asset (Pease et al. 2013: 4). When expenses need to be minimized, HRM practices that were approved without the support of financial measures are the first in line to be eliminated or reduced. One obvious reason behind this phenomenon is that HR professionals do not posses the analytic and data-based decision-making skills that are crucial to clarify and communicate the return of their investments. Therefore, it can be stated that it is important to pay attention to realizing the long-term benefits of people investments. (Murphy & Zandvakili 2000; Walsh et al. 2010.) “By insisting on the application of financial measures and customer data to the design and evaluation of its practices, HRM will become accountable for delivering results that are integrated with the overall business strategy” (Murphy & Zandvakili 2000). Walsh et al. (2010) agree by stating that in order for organizations to manage their investments strategically, they need to gather information and apply suitable analytics to support decision-making. As already stated, there exists great potential for HR to became a highly valued strategic partner in organizations. However, when comparing to other more strategically mature functions like finance or marketing, HR is still struggling. (Lawler, Levenson & Boudreau 2004; Vargas, Yurova, Ruppel, Tworoger & Greenwood 2018.) Walsh et al. (2010) argue, that decisions made in HR have heavy strategic implications but because of the intangible nature of these investments, it is difficult for decision makers to quantify these implications. Yet, these whims of HR outcomes are not unbeatable. When examining e.g. the function of marketing and its nature, many similarities to HR can be found. Like HR, marketing has to deal with uncertain decision paths, involvement of psychology, and unpredictable outcomes. Still, marketing has managed to adopt tools and analytics that help the decision makers to act according to the strategy. (Lawler, Levenson & Boudreau 2004; Walsh et al. 2010.) In fact, Boudreau & Ramstad (2007) point out that finance and marketing provide HR the blueprints to become influential decision-supporting function. Marketing, as well as finance, have both developed from professional practices (sales and accounting) into strategically important decision sciences. Thus, a pivotal solution for HR is to do the same and transform into a decision science: talentship. (Falletta 17 2014.) (See Figure 1.) According to Boudreau & Ramstad (2007: 19-20), this is simply about improving the decisions about the talents of people and how they organize and interact. However, Falletta (2014) states that, for HR, there is still a long way to go, as it is still very common in organizations to use information and data to support decisions that are already made. According to him, with the help of information and data companies should rather question the path they are on when it comes to HR strategy and planning. In other words, in order to really adopt an evidence-based approach, the data and information used in decision-making should be transformed into analytics, and further on into meaningful HR intelligence. (Falletta 2014.) Figure 1. From a professional practice into a decision science (Adapted from Boudreau & Ramstad 2004: 18.) HR intelligence is defined as “a proactive and systematic process for gathering, analyzing, communicating and using insightful HR research and analytics results to help 18 organizations achieve their strategic objectives” (Maldonado 2014). In other words, HR intelligence builds upon both scientific evidence and analytics, thus being supported by empirical as well as theoretical research. According to Falletta’s (2014) value chain (see Figure 2), HR intelligence can be seen as an opposite to intuition and prior experiences. As already stated, HR decisions have tended to lean to prior experiences and prior knowledge. Nowadays emphasis should be on basing decisions more on predictive analysis and scientific evidence. (e.g. Ulrich & Dulebohn 2015.) However, it is important to see HR intelligence both as a science and art, meaning that intuition and expertise should not be completely forgotten. (Fitz-enz 2010: 4-5; Falletta 2014; King 2016.) In addition, it is essential to mention that the concept of HR intelligence is used interchangeably with terms such as “evidence-based HR”, “HR decision science”, “workforce analytics”, “talent analytics”, and “HR analytics” (e.g. Fitz-enz 2010; Falletta 2014). Today, the most used term is arguably “HR analytics” and due to its popularity, it is used also in this paper. Figure 2. The HR intelligence value chain (Adapted from Falletta 2014.) As can be seen also from Figure 2., a mature decision science highlights the importance of data, information, and analysis and aligns them with its framework principles. Despite of the existing advanced technology, data availability, and the competences to distribute and report information, HR is still facing a huge challenge to master the creation of strategic insights that drive the organizational effectiveness. (Boudreau & Ramstad 2007: 37-38.) Next in this paper, in order to build a solid theoretical base for HR analytics, the concepts of data, measuring, and metrics will be discussed separately. 19 2.1.1. From data to big data The world has gone through significant changes as technological advancements have altered the way we live and work (Abdulmelike 2017). Organizations are expected to more and more handle and manage challenges concerning strategic capability as well as trends in how employees, stakeholders, and customers communicate and want to engage with one another. Aforementioned observations with continuosly increasing market and competitor demands, advocate the need for organizational changes. (Shah, Irani & Sharif 2017). Companies are, as a matter of fact, in increasing numbers, embracing the new world of smart working, business analytics, and increasing volume of data (Abdulmelike 2017). McAfee & Brynjolfsson (2012) even claim that data-driven decisions are simply better decisions. They agree with the ideas and opinions presented in the previous section of the paper by stating that data, and especially big data, helps managers to make decisions based on evidence rather than a hunch. (McAfee & Brynjolfsson 2012.) There are two ways to view data: structured and unstructured. Structured data refers to financial data while the term unstructured data is used when discussing economic or less tangible data. Examples of financial data are cash and other liquid resources, like stocks and bonds. Economic data, also called as off-balance sheet assets, on the contrary includes things such as market reputation, customer satisfaction, and community relations. These two forms of data are very distinct but yet they eventually blend with one another. That is because every economic data item should, at some point, turn into financial value. Financial value is reached when, for example, a stakeholder invests in the company stock, a customer acquires a product or service, or a talented employee applies for a position in the company. (Fitz-enz & Mattox 2014.) In todays world, technology is assisting to generate data so efficiently and fast that new words have to be invented in order to be able to describe the scale. Instead of megabytes and gigabytes, the storage capacity of data is nowadays expressed in exabytes and zettabytes. (Fitz-enz & Mattox 2014; Abdulmelike 2017.) This presence of large volume of data with huge variation generated in today’s digital environment is referred 20 as big data (Abdulmelike 2017). There exists multiple definitions for big data in the academic literature (Shah et al. 2017), but one often cited report defines it as “anything too large for typical database tools to be able to capture, store, manage and analyze” (Angrave, Charlwood, Kirkpatrick, Lawrence & Stuart 2016). Similarly, Grable & Lyons (2018) describe big-data as large data sets that are collected by organizations and governments. According to them, these data sets are so excessive that traditional data processing methods are not enough to make sense of the data. When analyzed in appropriate manner, “big data can provide more precise insights into hidden patterns, trends, and associations, especially in the context of human decision making”. (Grable & Lyons 2018.) Furthermore, big data can be defined in terms of volume, velocity, and variety. Volume refers to the total size of the data set, while velocity represents the speed at which the data is acquired and generated from both internal and external sources. Nowadays, due to the developments in computing power, the volume of collected data keeps on growing, as storing causes no limitations. Data is also gathered today in real time at a very fast pace. Finally, variety, or as Shah et al. (2017) call it, variability of data refers to the types of data being collected. Nowadays big data takes the form of e.g. social media messages, updates, and images, sensor readings, as well as phone GPS signals, while couple decades ago it was only basic demographics data, attitudes and opinions, and geographic information. (McAfee & Brynjolfsson 2012; Shah et al. 2017; Abdulmelike 2017; Grable & Lyons 2018.) In addition to the original three Vs, new ones have been added later on. A most discussed of them is veracity. Veracity, according to Grable & Lyons (2018), refers to the “noise” in the data. In other words, veracity presents “the varying credibility and reliability of different data sources” (Abdulmelike 2017). Simsek, Vaara, Paruchuri, Nadkarni & Shaw (2019) mention also vision, visibility, and value as considerable dimensions of big data. According to them, each dimension offers its own challenges, but also ways to conquer them, in accessing, storing, and using big data. For example, when it comes to the velocity dimension, issues such as transfer speed and storage 21 scalability arise. Veracity, in turn, is associated with matters related to uncertainty, authenticity, trustworthiness, and accountability. (Simsek et al. 2019.) As stated already, big data refers to a set of data that is too extensive to be managed and analyzed with basic database tools (King 2016). Capelli (2017), in turn, even argues that there is no such thing as big data in HR. According to him, in most organizations HR is still struggling to use data at all – not to mention big data. He explains himself by noting that the headcounts in companies are usually thousands, not millions, which literally does not fulfill the traditional definition of big data. (Capelli 2017.) However, recently the definition of big data has concentrated more on the smartness of data rather than the size of it. In other words, smartness of data refers to “the extent to which it is able to provide the material to conduct fine-grained analysis that successfully explains and predicts behavior and outcomes”. (King 2016; Angrave et al. 2016.) This new way of seeing and defining big data is arguably suitable in the field of HR as it embodies the data held in human resource information systems (HRIS). Data held on HRISs is considerably small by the standards of large unstructured data. However, it is big when viewed as quantitative data sets, and also able to produce smart insights due to its longitudinal nature. Therefore, it is possible to conclude that analytics include aspects from both the traditional as well as the updated interpretation. In other words, analytics consists of “traditional relational database and spreadsheet-based analysis, new forms of database software that allow very large quantities of data to be stored and organized more efficiently, and new techniques for representing and understanding data through visualization”. (Angrave et al. 2016.) However, the technical ways to fully integrate, organize, and analyze the data stored in traditional HRISs together with the data from the aforementioned large unstructured data sources is not yet established. Also topics such as privacy, consent, and ethics cause issues when dealing with HR related big data. (Angrave et al. 2016.) As a matter of fact, in most organizations HR’s data-driven procedures are still limited to means like collecting people metrics, exploiting comparison benchmarks, reporting to management, and getting rid of manual efforts by automating the reporting processes (HBR 2013). 22 Still, there is a lot organizations can achieve with already excisting relational data that is stored in HRISs. Today’s data-driven actions can be seen as crucial groundwork for more advanced and exact analytical use of big data and the enabler of data/evidence- based human capital decisions. Furthermore, in the future, when an organization is ready to move from basic descriptive data usage to more predictive modeling, it will have enough historical data to construct models on. (HBR 2013). According to Angrave et al. (2016) there are two central questions that should be considered in organizations when it comes to big data in the context of HR. Firstly, it is important to ask how analytics can be utilized to create, seeze, leverage, and protect the value within HR data. The second question, in turn, asks what are the ways and what is needed for excisting, essentially descriptive, analytics to evolve into programmes that are more strongly focused on measuring and modelling the strategic impact of workfore inputs in order to offer better decision tools for management. (Angrave et al. 2016.) 2.1.2. HR metrics and measuring Like already pointed out, as the importance of human capital in today’s business world is increasing at a rapid pace, there is a great opportunity for HR to become a core organizational function and participate in the development and implementation of the corporate strategy. However HR is still struggling to prove its worth and has failed to truly achieve the status of a strategic partner. One pivotal reason behind it is the fact that HR has not managed to acquire the needed level in analytic and data-based decision- making capability. This, in turn, stems from the lack of right and suitable metrics and analytic models. Unlike marketing and finance, HR tends to be unsuccessful in providing metrics that guide and assess HR processes and practices from a strategic viewpoint. (Lawler et al. 2004.) As a matter of fact, HR has been described as a soft function. The softness refers to the idea that people and metrics do not necessarily blend together. (Feather 2008; Tootell, Blackler, Toulson & Dewe 2009.) According to Dulebohn & Johnson (2013), all core business functions use metrics. Therefore, if HR wishes to consolidate its position as part of that group, metrics need to be applied. (Dulebohn & Johnson 2013.) This does not come as a surprise, given also 23 the unforeseen growth of human capital related data and different measuring and evaluation practices, stemming from the increasing interest towards people as organizational assets. Also the rapid development of HR systems, that have enabled the more efficient and quick data gathering and management, has supported this trend. To avoid risks associated with this kind of direction of development, HR must have the right frameworks applied to support and guide data analysis and decision-making. (Boudreau & Ramstad 2002.) Like Rasmussen & Ulrich (2015) point out, “rigorous analyses of loads of data on the wrong questions often have little practical value”. Mclver et al. (2018) say that it is crucial to be able to ask data the right questions and subsequently base the right metrics on the answers. Feather (2008) agrees when she says that HR has to ask itself what should be measured. According to her, without a measurement framework these huge amounts of collected data can become “an exercise in measurement for measurement’s sake”. (Feather 2008.) Boudreau & Ramstad (1998) are on a common ground as they note that HR measurement systems should systematically link to organization’s strategic goals and embody the “theory of the firm”, which refers to the linkage between people and organizational performance. Dulebohn & Johnson (2013) define a metric as “an accountability tool that enables the assessment of a function’s results”. According to them, in the field of HR the fundamental idea has been that through metrics HR can strengthen its partnership with other core functions by building a business case for their work. (Dulebohn & Johnson 2013.) Marler & Boudreau (2017) agree by characterizing HR metrics as “measures of key HRM outcomes”. In the academic literature, metrics are often theorized also through the different types of existing measures. Scholars have identified multiple categories for HR metrics, from which the most discussed ones are: efficiency, effectiveness, and impact metrics. (E.g. Lawler et al. 2004; Tootell et al. 2009; Dulebohn & Johnson 2013.) The efficiency metrics/measures concentrate to shed light on how well the HR function succeeds in its basic administrative tasks. In fact, until this day, most of the developed 24 HR metrics have been efficiency measures (Dulebohn & Johnson 2013). They focus heavily on productivity and cost, which arguably makes them the easiest kind of metric in terms of data collection. Good examples of efficiency metrics are, among others, cost per hire, days to fill a position, HR expenses per employee, and the percentage of performance reviews completed on time. (Lawler et al. 2004; Dulebohn & Johnson 2013; Carlson & Kavanagh 2018.) Efficiency metrics are operational in nature and therefore they alone are not enough. In order to measure the effectiveness, quality, and impact of HR operations on a larger scale, supplementary metrics are needed. (Dulebohn & Johnson 2013.) Lawler et al. (2004) agree by stating that many efficiency- focused metrics are limited, as they tend to ignore “the issues of service quality and impact of HR services on organizational effectiveness”. Effectiveness measures are less familiar to organizations than the ones belonging to the efficiency category (Tootell et al. 2009). HR effectiveness metrics (also called as “cost benefit metrics”) consider and measure whether HR programs and practices have managed to affect the targeted people or talent pools as intended. According to Dulebohn & Johnson (2013), these metrics typically include measures of the strategic skills and core competencies owned by the staff. Also, with the help of these metrics it is possible to assess how well and successfully critical positions are filled and what kind of actions and activities there are in place to develop talent. (Dulebohn & Johnson 2013.) Lawler et al. (2004) note that HR is often seen as responsible for acquiring, developing, and deploying talent in organizations. Through measures of talent quality, talent development, and talent deployment it is possible to monitor and assess how well the function has succeeded in its commitments. This is important also from the strategic perspective. An organization is destined to fail if its strategy has false assumptions about firm’s ability to fill critical positions and develop pivotal areas of expertise to support the strategy execution. (Lawler et al. 2004.) Impact is the third and highest level of HR metrics. This level is also called as strategic HR metrics (Dulebohn & Johnson 2013) and it explains in measurable terms how HR has managed to impact business outcomes in respect of finance, customers, processes, and people. In other words, in this context impact refers to the linkage between HR 25 operations and organization’s competitive advantage. These strategic HR metrics help to decide where to allocate and how to manage human capital resources in order to gain and sustain competitive advantage. According to Lawler et al. (2004), to be better able to guide these kinds of decisions, impact metrics tend to integrate HR data with other organizational data. HR function’s effect on operational effectiveness needs to be proved by demonstrating a connection between a specific HR metric and other metrics in the organization. (Lawler et al. 2004; Dulebohn & Johnson 2013.) Efficiency Effectiveness Impact Number of open requisitions Quality of hire Length of stay Number of hires Quality of the recruiting process Monthly productivity Hires by level Quality of service from recruiter Improved cycle time in job Hires by business unit Fit with the job Contributions to customer satisfaction Average cost per hire Fit with culture Contributions to quality Amount of coaching received Employee engagement Improved project quality International assignment Ability to coach Improved efficiency Mobile workforce Ability to be coached Increased cost savings Table 1. Examples of efficiency, effectiveness, and impact metrics (Adapted from Fitz- enz & Mattox 2014: 133-134.) In addition to these three aforementioned levels of HR metrics, Dulebohn & Johnson (2013) have defined also a fourth one: Human capital metrics. These metrics, as the name suggests, try to measure the value of human capital. The fact that organizations do not own their employees makes this a hard and difficult task. Workforce, unlike other capital assets, does not have precise documented purchase price or market value. Instead, the amount, type, and worth of human capital changes over time as people acquire new skills and get better in their areas of expertise. Given that the importance of 26 human capital in today’s business world is constantly increasing (Boudreau & Cascio 2017), it is no wonder that firms attempt to show their value in quantifiable terms. (Dulebohn & Johnson 2013.) Examples of human capital metrics are presented in the following table. (See Table 2.) Human capital metrics Expense factor Operating Expense _____________________________ Total Full-time Equivalent (FTE) Profit per employee Revenue − Operating Expense ___________________________ FTE Labor cost factor Compensation + Benefit Costs ___________________________ FTE Table 2. Examples of human capital metrics (Adapted from Dulebohn & Johnson 2013.) Even though there exists a consensus about the importance of HR metrics in transforming HR into a strategic partner, surprisingly many companies still admit using only efficiency metrics. In other words, in most cases companies are not even trying to develop HR effectiveness metrics, and even fewer are measuring the impact on the organization. Nevertheless, companies that assign resources to data gathering in order to apply metrics are arguably more likely to “develop predictive models that will contribute to sustained competitive advantage in managing and deploying their talent”. (Dulebohn & Johnson 2013.) It is important, however, to understand that metrics are just means of measuring something. The attention should be on that something instead of the metrics themselves. Nowadays, instead of just measuring the efficiency and effectiveness of the HR function, HRM should rather measure and figure out how much 27 value it produces for the overall business. (Ulrich & Dulebohn 2015.) Also, it is good to keep in mind that there still exists definitional ambiguity in the literature between the terms of HR analytics and HR metrics. According to Marler & Boudreau (2017) HR analytics are not HR measures or metrics, but rather “statistical techniques and experimental approaches” used to show and prove the outcomes of HR procedures. (Marler & Boudreau 2017.) 2.2. HR analytics Rest of the literature review delves into the main topic of the paper; HR analytics. Firstly, based on the already existing literature, the concept of HR analytics is defined and discussed. Also, the levels of HR analytics and analytical maturity are presented. Thereafter, the identified five key-factors effecting HR analytics maturity are each considered and discussed separately. Finally, as a result of the literature review, a framework for the study is formed. 2.2.1. Defining HR analytics Nowadays, in order to better understand their workforce and make wiser decisions, HR and business executives have understood the importance of data analytics. Even though accessing HR data has become increasingly easy in the last few decades, the HR analytics movement is still relatively new. (Subramanian 2017.) Marler & Boudreau (2017) challenge this notion by claiming that analytics in HRM has been around for years. According to them, notions of measurement in human resources can be tracked as far as the beginning of 1900s and the first book on the topic was published already in 1984. Therefore, it can be argued that the topic has been around for years but has gained a lot more attention in recent years. However, despite the huge interest around the human capital analytics, the majority of organizations nowadays find it challenging to evolve from operational reporting to analytics (Boudreau & Cascio 2017). 28 Analytics is stated to be a necessity for the HR function. Analytics acts as a tool for the function to create value from people and to strengthen its strategic influence in the organization. (Angrave et al. 2016.) Even so, no universal definition exists for HR analytics. In fact, according to Bassi (2011), the term “HR analytics” means distinct things to different people. Some people may see it as a process of systematic reporting using a selection of HR metrics while some believe “the only activities and/or processes that constitute HR analytics are those that involve “high-end” predictive modeling”. Both of these viewpoints are limited and present only separate components of HR analytics. (Bassi 2011.) Analytics can be seen to have developed in the crossroads of engineering, computer science, decision-making and quantitative methods. It is a discipline to organize, analyze, and understand the continuously increasing amounts of available data. (Angrave et al. 2016.) HR analytics, in turn, is a domain in the larger field of analytics applying analytic process in the organization’s human resource department. The aim is, in addition to gathering data on employee efficiency, to provide insights “into each processes by gathering data and then using it to make relevant decisions about to improve these processes”. (Chattopadhyay, Biswas & Mukherjee 2017.) Fitz-enz (2010) points out that analytics, like often assumed, is not all about statistics but rather it is a meeting point of art and science. The ideas behind dividing the field of analytics into two camps are that even though statistics play a major role in analytics, it is even more crucial to first understand the relations and interactions of the problem’s elements. “Analytics is first a mental framework, a logistical progression, and second a set of statistical operations.” (Fitz-enz 2010: 4; Ftz-enz & Mattox 2014: 3.) Baesens, De Winne & Sels (2017) agree as they diagnose that analytical HR models should do more than concentrate on statistical performance. According to them, two other important factors to consider are model interpretability and compliance. Interpretability means that every HR decision made using the help of analytics should be thoroughly argued and, if necessary, easily explained to parties involved. Compliance, in turn, highlights how important it is to use caution when interpreting analytical models. Also, gender equality and diversity are 29 dimensions that should be treated with respect when gathering and selecting data to be used in the analytical HR models. (Baesens, De Winne & Sels 2017.) Fitz-enz & Mattox (2014) further describe HR analytics as a communication device. They see it as way to bring together data from distinct sources, like surveys, records, and operations, to best describe the current situation and conditions as well as the probable futures. Thus, HR analytics can also be described as an evidence-based approach helping organizations to make better decisions. (Fitz-enz & Mattox 2014.) However, it is crucial to understand that analytics itself does not bring value to the function or the whole organization. Instead, analytics needs to be linked to business strategy to be valuable. (LaValle, Lesser, Shockley, Hopkins & Kruschwitz 2011.) Rasmussen & Ulrich (2015) are on common ground as they advice that analytics for the sake of analytics is not helpful. They bemoan that instead of focusing on the business challenges, analytics today tends to start with data. That contradicts with the fact that HR becomes successful by contributing to business decisions. Instead of just validating the already existing information and knowledge, HR should also try to act as an informant in business decisions. (Rasmussen & Ulrich 2015.) This links with the already discussed idea of moving from inside/outside perspective to outside/inside view (Ulrich & Dulebohn 2015). According to Rasmussen & Ulrich (2015), it is important for HR to move the focus from doing things right to doing the right things. Additionally, when trying to define HR analytics, the existence of multiple overlapping terms can cause confusion. For example, terms like workforce analytics, people analytics, and talent analytics appear in the literature and are often used interchangeably. (Van den Heuvel & Bondarouk 2016; McIver et al. 2018.) According to McIver et al. (2018), despite the term used, in the end it is about “the analysis of HR- related data, but also the integration of data from different internal functions and even data external to the firm”. Van den Heuvel & Boundarouk (2016) are somewhat in disagreement with them, as they believe that the differences in labeling go beyond basic semantics. In their opinion, for example the term HR analytics may suggest that it is the HR department who is mainly responsible for identifying and quantifying people drivers of business outcomes. The term “workforce analytics”, in turn, is somewhat 30 separated from the HR department and falls more on the shoulders of the whole organization. However, the term has a bit exploitative connotation to it. A more employee friendly alternative would be the term “people analytics”. (Van den Heuvel & Bondarouk 2016.) Nevertheless, in this paper the term “ HR analytics” is used, as it seems to be the most neutral option. Furthermore, this paper leans to the following definition of HR analytics: “The systematic identification and quantification of the people drivers of business outcomes, with the purpose to make better decisions.” (Van den Heuvel & Bondarouk 2016.) Even though, there has been a lot of discussion around the topic of HR analytics and people have identified the opportunities and benefits it can offer, organizations are still struggling to adopt it. In a study conducted by Deloitte (2015), HR analytics is, in fact, identified as one of the biggest challenges in HR practice nowadays. (Van den Heuvel & Bondarouk 2016.) The fact that workforce has gone through significant changes in recent times causes the need to manage it more efficiently. The staff in organizations today is more and more versatile covering many generations and continents. Still, it is customer, marketing, and financial analytics in organizations that are invested in. Firms should really ask themselves why the same approach has not been taken with their arguably largest and most crucial asset, workforce. Investing in HR analytics is not actually beneficial only to the HR function but serves other business areas as well. “By providing people analytics and insights to other areas of the business through an information loop, HR can help drive business value.” (Deloitte 2018.) Along with all the criticism, Edwards & Edwards (2019) have also seen development in the field of HR analytics. According to them, in recent years the attitudes towards HR analytics have changed considerably. Instead of being something new and attractive, HR analytics is nowadays considered to be almost a necessity in the offering of a HR function. (Edwards & Edwards 2019.) In fact, in a survey conducted by Deloitte (2018), 71 percent of participated executives identified HR analytics as important or very important. Unfortunately, organizations have not yet been able to show their conviction 31 in practice. Instead, firms are stuck in producing descriptive reports after another. There is no question that descriptive analytics would not be useful but the truth is that there are certain limitations. These descriptive reports only tell what is going on in an organization at a given moment without having anything to say about, for example, reasons behind the occurring events. Furthermore, descriptive analytics does not predict possible future outcomes at all, which is arguably the most notable factor differentiating predictive HR analytics from the analytics currently carried out by most organizations. (Edwards & Edwards 2019.) These different levels of HR analytics are presented and discussed next. 2.2.2. The levels of HR analytics HR analytics is divided into three different levels: descriptive, predictive, and prescriptive (e.g. Fitz-enz & Mattox 2014; King 2016). Descriptive analytics refers to the inspection of data and information in order to figure out what is going on in an organization at a given moment or what has happened in the past. In other words, it reveals and describes relationships and presents historical data patterns. (Fitz-enz & Mattox 2014; Sivarajah, Kamal, Irani & Weerakkody 2017.) According to Fitz-enz (2009), descriptive HR analytics offers also a way to drill down into the workforce to discover subgroups based on predetermined traits. He states that, this kind of workforce segmentation corresponds to customer segmentation done in the field of marketing. (Fitz-enz 2009.) Today, majority of organizations are still on the first level carrying out very basic descriptive HR analysis (Minbaeva 2017). Even though descriptive analytics brings less value to an organization than higher-level analytics, it is a prerequisite as it acts as a foundation for more advanced analytics. Examples of descriptive analytics are, among others, dashboards and scorecards, data mining for basic patterns, human capital segmentation, as well as periodic reports. (Fitz-enz & Mattox 2014.) The second level of analytics is the predictive analytics. There are three factors that have caused the predictive analytics to become a hot topic in HR. First of all, computing power has developed significantly and become more affordable. Secondly, because of cloud storages, more and more big data is available for HR to process. Finally, 32 nowadays there is a global war on talented employees, which is why organizations need to protect and pursue talent streams. (Mishra, Lama & Pal 2016.) Predictive analytics, unlike descriptive analytics, is about trying to determine the future with the help of current and historical facts (Sivarajah et al. 2017). So, predictive analytics gives meaning to the aforementioned patterns for future. It is about probabilities and potential impact. (Fitz-enz & Mattox 2014.) Predictive analytics includes statistical techniques, machine learning methods, and data mining models that are used to analyze and extract current and historical data and facts to make predictions about the future (Mishra et al. 2016). Examples of predictive analysis are decision trees, genetic algorithms, and neural networks (Watson 2015). As already noted, in predictive analytics the future is expressed with the help of probabilities. However, it is important to understand that these kinds of analytical predictions cannot be made with absolute certainty. It is possible to foretell the future only to some degree. (Fitz-enz 2009.) However, in order to achieve the highest possible degree of probability in forecasts, the following four things must be in place (Fitz-enz 2010). 1. A full comprehension of current and historical events. 2. An understanding of present trends and the drivers behind them 3. The ability to recognize consistency and change patterns 4. The availability of tools to describe the future probabilities The third and final level of analytics is prescriptive analytics. According to Fitz-enz & Mattox (2014), “prescriptive analytics goes beyond predictions and outlines decision options and workforce optimization”. So in other words, it predicts future outcomes based on complex data analyzes and provides decision alternatives and possible future outcome options. (Fitz-enz & Mattox 2014.) King (2016) describes prescriptive analytics as designing treatments to overcome currently occurring obstacles. Analysis at this level is very rare in HR (Fitz-enz & Mattox 2014). Even the organizations considered as pioneers in analytics have very seldom managed to master this advanced analytics (King 2016). 33 Figure 3. The levels of HR analytics (Adapted from Sivarajah et al. 2017.) In addition to the three, already presented, levels of analytics, Mishra et al. (2016) present two complementary levels: inquisitive and pre-emptive analytics. The level of inquisitive analytics comes after descriptive analytics and refers to the examination of data in order to approve or reject business propositions. Examples of inquisitive analytics are, among others, analytical data drilldown and statistical analysis. Pre- emptive analytics, in turn, takes place after prescriptive analytics and focuses on the capacity needed to take precautionary actions on issues that will possibly have an undesirable effect on organization’s performance. (Mishra et al. 2016.) Moreover, Bersin (2014) has developed his own model to view the levels of HR analytics. It is called “Talent analytics maturity model” and it has four distinct levels: operational reporting, proactive–advanced reporting, strategic analytics, and predictive analytics. Organizations belonging to the first level focus only on operational ad hoc reporting that reacts to business demands. Data isolation and difficulty of analysis are characteristics for that organization group. Proactive–advanced reporting, in turn, refers 34 to the state of analytics where operational reporting is used for benchmarking and decision making. Good examples are dashboards and multidimensional analysis. Third level is “strategic analytics” and it concentrates on segmentation, statistical analysis, and development of people models. The goal is to understand the cause and delivery of actionable solutions through the analysis of dimensions. The final level is labeled “predictive analytics” and on that level firms are developing and executing predictable models, scenario planning, as well as risk analysis and mitigation. On predictive analytics level, organizations have already managed, to some extent, integrate HR analytics to the overall strategic planning. (Bersin 2014; Isson & Harriot 2016.) Figure 4. Degrees of analysis and business intelligence (Fitz-enz 2010: 186) Like already mentioned, many organizations are still executing analytics on a very basic level in forms of basic operational (descriptive) metrics and reports (e.g. Boudreau & Cascio 2017; Minbaeva 2017). Unfortunately, data and metrics by them selves are not enough to help managers to make effective decisions (Fitz-enz 2010: 186). In fact, it can be even frustrating and confusing for managers to receive loads of reports and 35 documents without any real insights (Fitz-enz & Mattox 2014). Therefore, it is crucial for organizations to progress on analytics levels to statistical analysis and eventually to predictions and optimization. By increasing the degree of analysis, organizations are also building up their business intelligence. (Fitz-enz 2010.) (See Figure 4.) Fitz-enz (2010) describes the basic reporting as looking to the rearview mirror; it is limited to shed light into the things happened in the past. If an organization wants to understand outcomes and predict possibilities and events, more advanced analytics must be applied. The pressure and need to adopt HR analytics can also stem from external sources. Business is nowadays highly global and competitive, and in order to survive in an environment like that, forecasting, predictability, and advanced modeling become helpful. (Fitz-enz 2010.) But moving up from one level of analytics to another is not an easy or simple task or otherwise more companies would have adopted HR analytics. In fact, according to a survey conducted by Deloitte (2019), only 26 percent of respondent organizations report using HR analytics effectively. A research conducted by Deloitte (2017a) suggests that the focus has, as a matter of fact, shifted from the already presented more “technical” stages of maturity to organizational enablement, culture, and skills development. Therefore it is insightful to present one more model concentrating on the organizational maturity when it comes to HR analytics. This model does not ignore the technical maturity but rather recognizes the fact that companies are still progressing it alongside with organizational maturity. The four levels in this model are labeled as fragmented & unsupported, consolidating & building, accessible & utilized, and institutionalized & business-integrated and they are supposed to be additive from one level to the next one. (Deloitte 2017a.) The first level (Fragmented & Unsupported) refers to organizations that haven’t really realized the value of data and its role as a core asset in gaining competitive advantage. The fact that data is not been treated as a value driver suggests two things. Firstly, companies are lacking data governance, which in turn undermines the value of data and makes the use of it unreliable and risky. Secondly, these kinds of firms have a culture where hierarchy, precedence, experience, and intuition overrun data in decision making. 36 As well as most other companies, also fragmented and unsupported organizations collect employee data through e.g. HR systems and surveys. However, these data collecting practices are often reactive and compliance-oriented in nature. Instead of having a dedicated HR analytics team, level 1 organizations have only distinct roles concentrating on HR analytics. These people are working in silos and mainly focus on basic reporting per requests. Approximately 14% of organizations are on the first stage of HR analytics maturity. (Deloitte 2017a.) Figure 5. Maturity levels of HR analytics (Adapted from Deloitte 2017a.) Today most of the organizations are located on the second level of HR analytics maturity as approximately 69% of organizations identify themselves as consolidating and building. These organizations have understood that gathering, tracking, and analyzing workforce data is valuable and therefore intuition, precedence, and former experience have started to loose their power in decision making to data and metrics. As data is understood as the core component of analytics, Level 2 organizations are investing in data consolidation, accuracy, timeliness, privacy, and security. The gradual goal is to have consistent data definitions in use in the whole organization for the most 37 important HR metrics. The amount of sources, from which workforce related data is gathered, is more or less the same than on the first level. However, data is collected much more intentionally and promptly. At this level, organizations also strive to form a central HR analytics group by hiring and bringing together HR analytics oriented professionals from across the organization. (Deloitte 2017a.) The third level of analytical maturity is labeled as “Accessible & Utilized”. On that level organizations concentrate on ensuring that a considerably broader audience has the access to data and insights via automated dashboards, shared services, etc. That is possible due to the efforts made on the level 2. Because of the broadening audience, data and insights become more often used and embedded in key decisions concerning people and talent. Little by little they start to be present in regular work and everyday decision making. Also new tools, models, and approaches are experimented to study employee behavior. All of these aspects cause the need to educate the broader HR population further in basic data literacy skills, where to find what data, and how to use it. Moreover, one significant difference between the levels 2 and 3 is the mindset. On the third analytical maturity level the mindset has shifted from HR measuring HR to the already discussed outside/inside approach (Ulrich & Dulebohn 2015). Summarized, it can be stated that organizations on level 3 see considerable increase in investments relating to, for example, bigger analytics teams, more advanced analytics tools and technology, and third-party partnerships. About 15% of organizations have managed to reach this level of maturity. (Deloitte 2017a.) The fourth and final maturity stage is labeled as “Institutionalized & Business- integrated” and only 2% of organizations have conquered this level. On this level HR analytics is seen as a part of organizational DNA as it is used throughout the company in people and business related decisions. In other words, HR analytics is elemental in working and decision-making. Also the data literacy in the HR function is on a considerably high level. Level 4 organizations have a wide collection of tools and technology in use. With the help of e.g. AI-aided, cognitive, and robotic process automation tools it is possible to quicker gather and analyze larger volumes of data, both structured and unstructured, from multiple different sources and use statistical 38 methods. This leads to insights that are more progressive, accurate, and timely and these insights in turn are innovatively used to make the organization a better place to work. Instead of expanding the core HR analytics team in number, the team pursues to connect with different parts of business and diversify its skills and capabilities. Therefore, it is likely that the boundaries of the team become blurry and undefined. In other words, and also like Rasmussen & Ulrich (2015) state, HR analytics team may transform into “a cross-functional team that is responsible for data-driven decision-making across the organization in all people- and business-related matters”. (Deloitte 2017a.) In order to better understand what is needed to succeed in HR analytics and to move from the lower levels of analytics to the more mature and predictive analytics, the topic should be examined in sections. In fact, Isson & Harriott (2016: 53) claim that in order to gain excellence in HR analytics, prudent investments are required in people, processes, and technology. In addition to these three dimensions, also data and governance are regarded as analytical capability enablers (Deloitte 2011). Therefore, next in this paper all of these five interrelated dimensions are examined and discussed separately. 2.2.3. The people behind HR analytics The overall shortage of data scientists is prevalent especially in HR. A solution to that, in addition to using the already existing analytics resources, tools, and technologies, is to build a team or a function of analytics consisting of people with various backgrounds. (Isson & Harriot 2016.) According to Fitz-enz & Mattox (2014), it is important that the analytic team fits with the organizational culture and structure and is positioned in a way that enables the hiring, training, growing, and retaining of the needed talent. Positioning and structuring can be done in many different ways, but it is always crucial to keep in mind the vision, brand, and culture of the organization. Analytic function can be centralized, decentralized, or some kind of variation between these two alternatives. However, before concentrating on the structure, it is important to focus on the people and activities. Without the right and suitable people, structure is indifferent. (Fitz-enz & Mattox 2014: 36.) 39 Fitz-enz & Mattox (2014: 36-37) note that the priority is to find the people who have statistical skills but also the enthusiasm and intuitive skills to go beyond analysis. The variety in backgrounds also enables the team to view data from different viewpoints and eventually to derive even more remarkable insight from it (Isson & Harriot 2016). McIver et al. (2018) have identified four central capability areas in HR analytics: math and statistics, programming and database skills, domain knowledge including HR expertise and behavioral science, and communication and visualization. As it is almost impossible to find an individual possessing all of these skills and competencies, also McIver et al. (2018) urge to form HR analytics teams. Isson & Harriot (2016) agree with the aforementioned skills and give examples of positions a skill-diverse analytic team might include. First off, they mention technical specialists who are responsible for working closely with IT teams to make sure “the collaboration and support needed to gather data from multiple sources and integrate, standardize, and govern it” works. (Isson & Harriot 2016.) HR analytics teams need also statisticians, data scientists, and business intelligence specialist who help organizations to gain deeper knowledge about their talent life cycles and to better understand what is going on and why. These individuals are skillful in numerical analyses and capable of offering different views on challenges and the ways to overcome them. Moreover, they are able to use predictive models in order to prepare for workforce planning needs. These needs are among others the identification of talents the organization is at risk of losing and the candidates most likely to be successful once hired. Last, but not least, Isson & Harriot (2016) identify business analysts and navigators as essential actors in analytic teams. This group of people can be seen as the contact between the HR analytic team and the rest of the business. That is because they are the ones who paint the pictures of possible better futures that play a huge role in winning the C-suite over. (Isson & Harriot 2016.) McIver et al. (2018) state that organizations usually need to look outside HR when building the HR analytic team with necessary skills. The fact is that most people in the HR position do not understand big data or analytics (Angrave et al. 2016). Ulrich & Dulebohn (2015) even state that many HR professionals chose the field of HR in order 40 to avoid working with numbers. Therefore, individuals with the desired capabilities could be found from elsewhere, like IT and finance departments. Borrowing or buying capabilities from the external market is also a possibility. (McIver et al. 2018.) Rasmussen & Ulrich (2015) take the discussion even further and drastically state that HR analytics should be taken out of HR. They base their statement on the idea that when HR analytics mature, it first starts to collaborate more with other functions and departments and eventually end up looking at human capital elements in the whole value chain. In other words, it becomes a part of cross-functional analytics. (Rasmussen & Ulrich 2015.) Through analytics it is possible to yield truly new insights only if several distinct fields and perspectives are combined together. Therefore, HR analytics must go beyond HR issues and join the bigger cross-functional business analytics. Rasmussen & Ulrich (2015) claim that adding a denomination in front of the word “analytics” indicates immaturity and unpreparedness to be part of the larger concept of analytics. According to them, HR analytics generally needs a couple more years to mature within the HR function before being ready to give away the prefix. This process can be sped up through ”importing business analytics talent to run HR analytics”. In fact, in most cases it is easier to teach analytics professionals HR than to shape HR professionals into statistics and analytics experts. (Rasmussen & Ulrich 2015.) Bassi (2011) disagrees with the aforementioned ideas as she says that no other function than HR should take the lead in HR analytics. Instead, HR should develop its capabilities and capacities to answer the new business requirement. (Bassi 2011.) When it comes to constructing strong HR analytics capabilities, organizations normally tend to concentrate only on strengthening and developing the skills of the core analytics team. Even though that is a crucial aim, the skills among rest of the HR population should not be forgotten. It is equally important to enhance the data-literacy skills of analytics end-users. (Deloitte 2017a.) Sometimes the responsibility for educating and raising the level of analytical awareness within the HR function falls upon the analytics team itself (Boudreau & Ramstad 2004). Data literacy skills refer to simple data fluency, meaning the familiarity with basic statistical concepts, comprehension of the 41 difference between correlation and causation or practical and statistical importance, ability to slice and dice data based on uncomplicated parameters, and understanding of appropriate data sources and formats. It is central to understand that the goal is not to have all HR professionals able to perform complicated analytical tasks. Instead, the idea is to increase the level analytical mind-set and data literacy. (Delotte 2017.) There are multiple reasons why it is so important to be able to think analytically. First of all, when a people related problem arises, data literacy makes HR professionals more likely to ask the right questions. Moreover, being skillful in data literacy helps to better understand the problem at hand, instead of just jumping straight to the “solution mode”, which is very typical for HR. Lastly, HR professionals with a basic understanding of analytics are better equipped to distribute insights, created by the core HR analytics team, to the rest of the business. (Deloitte 2017a.) According Subramanian (2017), organizations today are far from the desired level when it comes to analytical abilities. He says that analytical abilities must be strengthened throughout organizations, but HR function requires extra attention. These analytical capabilities can be increased by hiring, but as there is a prevailing shortage of qualified data scientists with human capital knowledge (Subramanian 2017; Kryscynski, Reeves, Stice-Lusvardi, Ulrich, Russell 2018) in the market, a better and more effective option would be to train existing staff. (Subramanian 2017.) A study conducted by Deloitte (2017a) suggests that the size of an analytics team does not differ significantly between mature and immature organizations. Analytically mature organizations have approximately 6-10 members in their core analytics team while in immature companies the number is 1-5. However, other correlating factors can be identified. Instead of the size of the analytics team, there are three crucial elements that link with improved maturity. The first one is a team structure that enables strong connections with the business. It is important that teams have dedicated roles and responsibilities in order to “investigate the analytics needs of the business, communicate those needs to the core analytics team, and provide intelligence from that team back to the business”. Secondly, as already discussed, teams must have a diverse set of 42 competencies, skills, and expertise. Finally, what differentiates mature teams is that they report directly to the Chief Human Resource Officer (CHRO). (Deloitte 2017a.) 2.2.4. The processes of HR analytics The second key factor in creating high-impact analytics is the processes (Isson & Harriot 2016). According to Fitz-enz & Mattox (2014), analytics is all about understanding the interrelationships and interdependencies of different parts of a process. Processes consist of inputs, throughputs, and outputs and by understanding the correlations between them it is possible to intervene to the two aforementioned to improve the outputs. (Fitz-enz & Mattox 2014: 50-52.) It can be argued that sometimes organizations tend to focus too much on the data. Data alone is not enough to bring value to an organization – actionable insights are also needed. For that, Isson & Harriot (2016) have developed the IMPACT cycle. (See Figure 6.) IMPACT cycle is to “guide analysts through the process of ensuring they are insightful business partners, rather than just purveyors of data”. (Isson & Harriot 2016.) Before presenting the different stages in an analytics process, it is necessary to discuss the needed building blocks behind it. McIver et al. (2018) state that an agile HR analytics process is built upon a foundation of three crucial components: workforce analytics capability, workforce analytics vision, and strategic HRM perspective. The first building block, the people and their analytical capabilities, is already discussed in this paper but the other two blocks still require examination. Workforce analytics vision provides two important aspects. First, it acts as a compass setting the direction and focus for the people behind the HR analytics. Secondly, it creates transparency by communicating the goals of HR analytics for the whole organization. In other words, vision shows the directions and helps the whole organization to understand what the people behind HR analytics are striving to do. Communication around the vision emphasizes themes like data-driven insights and expected performance benefits. (McIver et al. 2018.) In the long run, this kind of open communicating arguably leads to the development of data-driven decision-making culture. In fact, it has been claimed 43 that “organizations can only reach their full potential in analytics maturity when data- driven decision-making is embedded in the culture”. (Deloitte 2017a.) Figure 6. The IMPACT Cycle (Isson & Harriot 2016: 56.) Strategic HRM perspective is the third required building block behind successful analytics processes. A strategic HRM framework or roadmap is a necessity if an organization desires to fully benefit from the potential of HR analytics. That is because these kinds or frameworks and roadmaps focus the attention on important outcomes and elements behind these results. They outline the central variables in a way that gives suggestions about their linkages that could be examined through HR analytics. Strategic HRM frameworks are also very useful for the people and teams behind HR analytics as they guide them in building long-term strategic capability as well as to continuously make operational improvements. Moreover, these kinds of roadmaps lay out a more 44 comprehensive picture of the direction and offer a way “to create user buy-in through more short-term immediate impact projects”. (McIver et al. 2018.) A study conducted by Deloitte (2017a) agrees by claiming that strategic alignment with the business is key in analytical maturing. Once these three building blocks are in order, organizations have better chances to be successful in their HR analytics processes (McIver et al. 2018). An organized process consists of all the steps needed to provide answers (Fitz-enz & Mattox 2014). The first step is naturally to identify the critical business questions that need to be answered (Isson & Harriot 2016). Like already noted, it is important for organizations to be able to ask the right questions (e.g. Rasmussen & Ulrich 2015; Simsek et al. 2019). While forming these questions, HR function must have an outside/inside approach that connects them with the broader business context (Ulrich & Dulebohn 2015). Second step is mastering data and it refers to the assembly, analysis, and synthesis of the available information. While conducting these actions, analytics team must keep in mind the already set questions. (Isson & Harriot 2016.) Like already said, HR analytics team must include people with wide range of background and skills (e.g. McIver et al. 2018) in order to produce simple, clear, and comprehensible visual presentations from the data available. These presentations are in forms of charts, graphs, tables, interactive data environments etc. (Isson & Harriot 2016.) Third step in the process is labeled as “Provide the meaning”. After producing visual presentations, analytics team must articulate clear and succinct interpretations of the data and visuals. Once again this is done in the context of the critical business questions formed at the beginning of the process. (Isson & Harriot 2016.) Analytics for the sake of analytics is the wrong approach. Instead of just validating existing knowledge in practice, HR should try to add value to business decisions. (Rasmussen & Ulrich 2015.) The next step is to provide actionable recommendations based on the interpretations (Isson & Harriot 2016). Also Kryscynski et al. (2018) identify the importance of being able to translate results into understandable and actionable insights. As it is easier to react to a suggestion than to create one, it does not matter if they are off base. Furthermore, whenever possible, add a rough monetary figure “to any revenue 45 improvements or cost savings associated with your recommendations”. (Isson & Harriot 2016.) Fifth step is communication. According to Barrette (2015), as much as 45% of analytics team’s focus should be on communication. Also LaValle et al (2011) have identified the importance of communication in turning insights into value. Analytics team should have a multipronged communication strategy that guarantees insights to flow as far and as wide as possible in the organization. Communication can have several different forms like an interactive tool or an executive memo. Lastly, it is time to track outcomes. Analytics team should establish a way to follow the impacts of their insights. Future follow-ups with business partners are especially important. That way analytics team can keep track on what has been done, what were the impacts, and have new critical questions possibly risen during the process. (Isson & Harriot 2016.) 2.2.5. Technology in HR analytics Marler & Boudreau (2017) define HR analytics as: “a HR practice enabled by information technology that uses descriptive, visual, and statistical analyses of data related to HR processes, human capital, organizational performance, and external economic benchmarks to establish business impact and enable data-driven decision- making”. Therefore, it can be argued that technology plays a big part in becoming analytically mature. As a matter of fact, it has been researched that strong analytics teams concentrate on actionable and scalable information delivery solutions. It is crucial to be able to deliver the insights to a wider audience, to every decision, and more frequently. In low maturity organizations this tends to be on the shoulders of the analytics team. Worst-case scenario is an organization without a team of analytics where the responsibility falls on a single individual. (Deloitte 2017a.) Nowadays, the HR function is using cloud-based technology, software as a service (SaaS), and client-server solutions, while few decades ago they were still dependent on mainframe computers and manual payroll systems. The emergence of talent management systems has made it possible for organizations to gather, store, and 46 manage a wide assortment of employee and HR data. While the amount of data increases, more and more technology companies with solutions to master the data appear on the market. (Isson & Harriot 2016.) Jones (2015) agrees, as she says that the amount of HR and talent applications and standalone analytics solutions have increased quickly. According to her, today those solutions and products that do not advertise the readiness for analytics are scarce. Instead, solutions include reporting tools, which are in most cases incorporated into human resources information system (HRIS) and talent management applications. (Jones 2015.) HRIS and talent management software brings together data from a wide range of already existing HR-related databases into a single cloud-based data warehouse. Through HRIS software packages it is possible to e.g. assist career planning, manage performance and learning enrollments, as well as share knowledge in a user-friendly way. The main purpose of HRIS is to enhance HR processes and operations by making it more effortless and faster to access and comprehend key HR and employee data. In addition to these usual features, “all the major integrated talent management suites also include analytics modules”. They are displayed as a key advantage compared to the older HRISs. (Angrave et al. 2016.) However, there are still deficiencies when it comes to HRISs. As already stated in this paper, HR analytics literature emphasizes that HR data is mainly realized by using it to answer critical strategic questions about how people create value for the company, so that value can be apprehended and exploited (e.g. Rasmussen & Ulrich 2015; Angrave et al. 2016). “The analytics modules of HRIS software packages as they are typically sold and implemented do not have the capacity to perform this sort of analysis.” Even the big multinational corporations (MNCs) that have spent enormous amounts of money and resources in HR analytics and have achieved results in analytics in other areas of business admit that their HR analytics programs are stuck in reporting historical information only. In other words, many organizations have failed to develop forward- looking strategic analysis with very little ideas how to implement big data into their HR analytics programs. (Angrave et al. 2016.) 47 According to Bersin (2015), the reason why existing talent management software is already outdated is the fact that it was designed to bring solutions to the problems of the last decade. Nowadays, instead of the “war for talent”, organizations are facing challenges relating to engagement, empowerment, and environment. Seeing people entirely as “talent” has become a narrow and limiting concept. Organizations naturally want to hire, train, develop, and lead people in order to gain results but it is important to see employees as individual people who have their own reasons and motives to come to work. Therefore, the focus should rather be on “people” than “talent”. (Bersin 2015.) Sommer (2015) gives a contrast to Bersin’s (2015) ideas as he claims that instead of HR technology, the problem is HR itself. According to him, HR departments are not ready for new technologies at the moment. That is because of the lack of awareness or skills in these technologies. Like already mentioned, HR function is missing the people with analytical competencies. (Sommer 2015; DiClaudio 2019.) Therefore, a rebalancing of talent within HR organizations is necessary. “New skills, capabilities and insights are needed to make HR more relevant and able to exploit today’s new HR technologies.” (Sommer 2015.) Nevertheless, HR analytics can be seen as a journey and therefore it is wise to start small and first master the workforce information already at disposal. Also, it is good to leverage the experts in the field of traditional analytics technology as they can help organizations to map out the best and most suitable approaches. (Isson & Harriot 2016.) According to Berry (2015) many organizations tend to spend a lot of money in building their own technology infrastructures while, in fact, “rightly positioned and properly executed, partnerships with vendors will pay a multiplier several times greater than the return on investment in your own technology”. Truth is that today’s tech vendors are very service minded and want to help their clients to over come their challenges. By choosing to go with a vendor, HR analytics can also avoid being considered as a huge expense that can end up on chopping board when cutting costs. (Berry 2015.) 48 2.2.6. Data in HR analytics Analytics can draw insights from both internal as well as external data sources (Deloitte 2011). Examples of data sources are among others HRIS, learning management systems (LMS), social media, surveys, and public data from outside the organization. For every data source, it is important to be aware, what the data is, where the data is housed (DiClaudio 2019.), who owns the data, how to collaborate with the owner, how to access and integrate the information with other data, and how valid and reliable the data is. (Pease et al. 2013.) However, like already referred, the technical ways to integrate, organize and analyze data stored in traditional HRIS with data from these larger unstructured sources are not yet fully established. Privacy, consent, and ethics are also issues that need to be considered when storing and analyzing HR and people data. (Angrave et al. 2016.) Once you have a vision of what you need in order to create reports and conduct analysis, the next step is to determine where the critical data resides and who owns it. According to Fitz-enz & Mattox (2014: 74-75), in most cases it is IT that controls the data. That is because in the majority of organizations the tools, like LMS or HRIS, are implemented and maintained by IT. (Fitz-enz & Mattox 2014.) Due to the advancements across the technology industry, it is nowadays much easier to get a hold of information as various types of data are nowadays resident in one system. Before organizations started to invest in HRIS upgrades the data was held in separate pieces of software designed to perform different HR processes. (Fitz-enz & Mattox 2014; Angrave et al. 2016.) Another supporting factor is the fact that distinct systems today are often integrated under one single service provider. One good example is SAP, that acquired SuccessFactors, and is now the owner of a system that can report hiring information, training histories, compliance, performance appraisal scores, high potential status, promotion history, compensation and benefits information etc. However, there are still organizations that house their valuable HR data in multiple distinct systems. (Fitz-enz & Mattox 2014.) 49 After identifying the source of data, the next step is to know who has the ownership of the data. According to Fitz-enz & Mattox (2014), the owners are sometimes unable or unwilling to share the data. There is often an acceptable reason behind that. The reason can be, for example, that it is not possible to export data to a file from the system. Another, and a more common, reason is the sensitivity of data, which is due to government regulations or organizational policy. This kind of data includes personal information, such as gender, age, ethnicity, and medical history and is therefore protected. (Fitz-enz & Mattox 2014.) The developments in technology have offered considerable advantages in terms of data processing efficiency and productivity. However, at the same time these advancements cause worry about individual rights. There is discussion about whether these otherwise positive developments have an adverse effect on the privacy of individuals and that this would be aggravated when viewing in an international context. These concerns have led to measures like the enactment of the General Data Protection Regulation (GDPR). (Ustaran 2018.) Besides privacy, the ownership of data is nowadays, in fact, a complicated issue. Due to the existence of, for example, social media, there are tons of real time data available. (Sivarajah et al. 2017.) Moreover, data can be gathered on e.g. what the employee does, who they communicate with, and what they communicate about. That kind of information resides in the location data from cell phones, internet browsing histories, electronic calendars, email, phone records, SMS messages etc. However, there are important issues of privacy, consent, and ethics to consider when housing and analyzing HR and people data. (Angrave et al. 2016.) With advances in privacy regulations, such as the already mentioned GDPR, HR analytics practitioners may have to re-think their approaches and procedures. Although lawyers, ethicists, and management scientist are already focusing on these issues, it still remains a blind spot for the HR analytics profession itself and therefore requires attention. (Tursunbayeva, Di Lauro & Pagliari 2018.) Alongside the issue of ownership, arise the questions about controlling and ensuring data accuracy. (Sivarajah et al. 2017.) To any analysis it is essential to have quality data. Before engaging with analysis it is necessary to examine the data in order to be sure it contains the appropriate measures 50 and that those measures are collected and stored consistently. (Fitz-enz & Mattox 2014: 80.) In fact, Andersen (2017) states that bad data is one of the biggest problems when considering the quality of HR analytics. According to him, low-quality HR data cost greatly for organizations. Therefore, in order to be able to guarantee better data quality and avoid additional costs, it is crucial to understand why bad data happens. The first reason is the lack of coherent data strategy. In other words, purely operational approach to data is not enough to ensure proper data quality. Another mistake is to assume that analytics software is the answer to the problem while in reality it is just another data collection tool. “To get the most out of HR analytics you must go through a strategic data process and decide what data are of strategic importance to you and how they ideally look like.” Thirdly, it is crucial to acknowledge that the data are as good as the component inputs – “garbage in, garbage out”. Finally, not having critical data sources affects the data quality. All in all, analysis and decision-making based on poor and low- quality data is meaningless. (Andersen 2017.) 2.2.7. Governance in HR analytics In this paper the last identified factor to have an influence on HR analytics maturity is governance. It is important for organizations to identify how analytical decisions are made and who should be the accountable for facilitating the analysis and leveraging it insights. (Deloitte 2011.) According to Feinzig (2015), once the analytics objectives are established it is central to communicate expectations and define working relationships with stakeholders. Like already stated, HR professionals pursue to make sure that the work of HR function contributes positively to overall business outcomes. Therefore, it is necessary to cooperate with business leaders and HR business partners in order to single out the business challenges and opportunities that can be addressed through workforce related actions. Furthermore, the members of the analytics team have to have a proper understanding of the key metrics used in managing the business and their current levels. HR business partners are helpful in identifying, which HR related actions impact the business metrics. (Feinzig 2015.) 51 It is arguably possible to identify data governance from the more comprehensive concept of governance. According to Patre (2016), proper data governance ensures that people use analytics as it was intended. The lack of data governance not only undermines the value of data but also makes it risky to use in any shape or form (Deloitte 2017a). In an efficiently functioning analytics process, the data inputs have to be consistent and reliable in order for information outputs to be relevant and comparable. When describing computing information, the four desired adjectives are relevant, reliable, comparable, and consistent. Effective governance implies that data used in decision-making is of consistent quality and from trustworthy sources. Moreover, when being efficient, “governance leverages connectivity and technology to enable the comparison of data from many different sources and to deliver relevant analysis.” (Smith & Heffernan 2019.) The data governance problems can be divided into two camps. First of all data governance can be ineffective, which means that the data quality is insufficient to be used in decision-making. Inefficient data governance, in turn, refers to data that cannot be accessed. All of the governance issues cause people not to trust the data and the decisions made based on it. Usually the mistrust is due to ineffective governance. It is possible that organizations are unaware of critical governance tools or have gap in communication between the data owners and the end users. Ineffective governance is normally easier to resolve, as inefficiency is sometimes needed due to regulations, like the already mentioned GDPR. (Smith & Heffernan 2019.) Feinzig (2015) emphasizes that it is crucial to understand data privacy legislations in each country where data is being gathered and analyzed. According to her, these laws and rules must be followed at all times. (Feinzig 2015.) Analytically mature organizations tend to ensure effective and efficient governance by establishing an organization wide data council or making sure the already existing data council has enough HR representatives. According to study conducted by Deloitte (2017a), data councils normally consist of couple of individuals from each business unit and corporate function. (Deloitte 2017a.) 52 2.2.8. Theoretical framework of the study This paper investigates the factors affecting HR analytics maturity. The aim is to figure out the current status of HR analytics in the case company as well as to identify the hindering factors in its development process. According to the already existing literature, there are arguably five main topic areas that are central in HR analytics: people, processes, technology, data, and governance. Furthermore, within these topics it is possible to identify more specific hindering factors. All of these together form the following theoretical framework of the study. Figure 7. Theoretical framework of the study 53 3. METHODOLOGY In this part of the paper the specific procedures and techniques used to identify, select, pr