The HR analytics cycle: a seven-step process for building evidence-based and ethical HR analytics capabilities

Step 1: determine stakeholder requirements

Determining stakeholder requirements is vital to the overall success of any HR research and analytics initiative. It is much more than meeting with a few influential or vocal stakeholders each year to formulate the annual HR research and analytics agenda. It is about establishing and cultivating a partnership and becoming a legitimate player by adding value to the business. With respect to the HR analytics cycle, an ongoing and proactive partnership with key stakeholders is an essential role to ensure trust and upfront legitimacy to obtain an accurate picture of the most pressing organizational problems.

Who are the stakeholders? A stakeholder in the broadest sense is anyone who is directly or indirectly affected by HR analytics work. Stakeholders in the context of HR analytics include executives, line managers, senior HR leaders, HR business partners, employees and, in some cases, human capital technology vendors. Each stakeholder has a different perspective and set of concerns regarding HR analytics practices and activities. Whereas line managers are usually most interested in the data visualization and reporting of key metrics and analytical insights, executives and senior HR leaders are generally more interested in how HR analytics enables HR strategy and execution, critical workforce decisions and other important business outcomes.

During this step, many interesting questions are likely to arise. Carl Sagan, in his book, The Demon-Haunted World: Science as a Candle in the Dark, explains “There are naive questions, tedious questions, ill-phrased questions, questions put after inadequate self-criticism. But every question is a cry to understand the world. There is no such thing as a dumb question” (Sagan, 1996, p. 323). While this is a very popular axiom and largely true, we've encountered a fair share of unusual and potentially unethical questions by some well-intended stakeholders in our work. To be fair, well-intended stakeholders are also bombarded with the latest management fads and trends (Barends and Rousseau, 2018; Rice and Boudreau, 2015). Hence, those leading HR analytics in organizations have an important role and responsibility in assisting stakeholders with framing HR research and analytics questions that are realistic, measurable, valued added and more importantly – evidence-based and ethical. Determining stakeholders' requirements is important to formulate and frame HR research and analytics questions, identify strategic and tactical HR research and analytics priorities, secure stakeholder commitment and support and provide communication on the ongoing progress of key HR research and analytics initiatives.

Step 2: define HR research and analytics agenda

Once the stakeholder needs and expectations are identified, it is time to define the HR research and analytics agenda. An HR research and analytics agenda may be long term or short term. The constantly evolving nature of business and the “future of work” are redefining what is considered long term versus short term. In our age of on-demand data aggregation and visualization, algorithms, artificial intelligence and automation – long term is no longer 3–5 years out. One year is considered the long-term norm currently in virtually all industries. Conversely, short-term requirements tend to coincide with the organization's quarterly results, sometimes monthly. It important to note that short term doesn't necessarily mean tactical or reactive, nor is long term equated with strategic. Short-term and long-term research requirements can be both strategic and tactical in nature.

Tips for establishing the HR analytics and research agenda

1. Organize the general stakeholder requirements by theme or major topic;

2. Pose broad research questions for each theme or major topic and use stakeholders' language and terminology to the extent possible;

3. Under each of the broad research questions, begin generating more targeted questions and hypotheses that lend themselves to measurement and consider the ethicality of the questions and hypotheses;

4. Identify both the long-term and short-term requirements of the overall HR research and analytics agenda;

5. Share the HR research and analytics agenda with key stakeholders and go through an iterative process of refinement; and

6. Lastly, strive for a balanced HR research and analytics agenda in terms of reactive and proactive work. A relatively recent HR analytics study conducted across Fortune 1,000 firms revealed that on average nearly 40% of HR analytics priorities were determined by the HR research and analytics team while approximately 60% were driven by stakeholders (Falletta, 2014). Hence, the HR analytics team should devote some resources to research projects that are strategic in nature, aside and apart from what was generated during the stakeholder requirements step.

Step 3: identify data sources

Once the HR research and analytics agenda is established, the next step is to identify the sources of data that will help to answer the HR research and analytics questions and hypotheses. Data sources may be either public or private. Public data reside in university libraries and governmental databases (e.g. UK's Office of National Statistics, U.S. Department of Labor Statistics). Private data includes an organization's internal employee data housed in its HRIS as well as external benchmarking data from “best-in-class” organizations. Research reports and results gathered by membership-based consortia (e.g. Global Centre for Work-Applied Learning, Centre for Evidence-Based Management, Gartner CEB, The Conference Board and the i4CP) and academic think tanks (e.g. Institute for Employment Studies in the UK, Cornell's Center for Advanced Human Resource Studies, University of Southern California's Center for Effective Organizations) are credible sources of private data and information. Sources of data may or may not exist depending on your organization's current HR research and analytics practices.

While reviewing data sources (refer to Table 3), questions may arise as to whether the organization's HR research and analytics practices are still useful to the business (e.g. some practices may have become institutionalized over the years). Therefore, some tough decisions may need to be made with respect to modifying existing HR research and analytics practices, discontinuing outmoded or symbolic practices and adopting new ones.

In terms of adopting new HR analytics practices and solutions (refer to Table 4), we caution organizations to do their homework before plowing into uncharted territory or chasing the next human capital management (HCM) technology or artificial intelligence (AI) platform. Too often, HR practitioners are too enamored with HCM technologies designed to select, manage, measure, track, monitor, quantify, nudge, engage and coach people in the workplace. These novel technologies, while promising, typically rely on a proprietary algorithm and unexplainable “black box” without fully understanding the underlying mechanisms at work (i.e. showing which factors or reasons led to human capital prediction or decision and how). According to Angrave et al. (2016), analytics must be rooted in an understanding of the data to be used and the context under which data were collected if any meaningful insight is to be gained.

In our view, if an HCM technology vendor is unwilling to crack open the proverbial “black box” – don't work with them. Although HR analytics does involve the use of technology to collect, manipulate and report data (Marler and Boudreau, 2017), it is critically important to weigh the potential rewards and risks including the legal and ethical implications associated with various HCM technologies.

Step 4: gather data

This step of the HR analytics cycle involves the actual collection of data through primary research, secondary research or mining the organization's internal data in the HRIS. Primary research is new or original research that addresses a specific research question or set of questions (e.g. a research project or experiment to identify which factors enable or inhibit employee engagement and performance, selection research using validated personality instruments for leadership succession, employee and organizational surveys for action planning and change, organizational network analysis to determine the level of collaboration by specific jobs and roles). Primary research can be done in-house if an organization has the HR research and analytics capabilities or in partnership with credible think tanks or universities (Simón and Ferreiro, 2018). Secondary research is data and information available through existing sources (e.g. literature review of journal articles and reports from credible institutes, HR benchmarking, labor market databases). The most reliable and trustworthy secondary sources of information for evidence-based decisions and practice come from scientific research studies (e.g. systematic reviews, meta-analyses, literature reviews) (Briner and Barends, 2016). Theory-driven (deductive) and data-driven (inductive) approaches to mining and modeling data from the organization's HRIS, SaaS-based platforms (e.g. data aggregation and visualization products, employee engagement tools, social or organizational network technologies) and other external data sources are another way to gather, query and analyze data about the workforce, provided it is done ethically and responsibility.

Step 5: transform data

Transforming data into useful and meaningful insights is arguably the most important, yet most challenging, step. Many advancements and innovations have been made by leading-edge software firms (e.g. Oracle, Salesforce, SAP, SAS, Workday) that incorporated HR analytical capabilities within their suite of products such as predictive analytics, process analytics, text and sentiment analytics and real-time analytics, to name a few (Strohmeier et al., 2015). Similarly, several promising data aggregation and visualization platforms have entered the market. However, none of these enterprise platforms and SaaS-based tools can magically codify, analyze, visualize and interpret all the disparate “Big Data” at our disposal (Angrave et al., 2016). This is especially true for evidence-based findings and insights derived through scholarly research as well as applied HR research and experimentation projects performed in organizations.

In the context of HR strategy, much of this work is still done manually by qualified HR researchers, analysts and data scientists (Falletta, 2014). It is suggested that organizations start small and build their HR analytical capabilities overtime. Organizations should perform a meta-analysis (i.e. an analysis of analysis) across a handful of targeted data sources. We refer to the term meta-analysis loosely and not in the traditional sense in which scholars use advanced statistical procedures to combine the results and examine the effect sizes of multiple scientific studies (Barends and Rousseau, 2018). Therefore, meta-analysis in the context of HR analytics is a practical approach to explore and understand multiple data sources in relation to each other. For example, to what extent are the results from individual 360-degree assessments consistent with your employee survey data, exit survey data, or actual turnover? Are high-potential, emerging leaders leaving the organization for the same reasons year over year (e.g. little to no advancement and promotion opportunities, few organizational leadership opportunities, lack of decision rights, low base pay relative to the market)? Performing a meta-analysis enables you to answer these questions and, more importantly, codify and make sense of disparate data sources to glean critical workforce insights. Performing the meta-analysis can be simple or complex. This largely depends on the nature of the data gathered, sophistication and competency of the HR researcher, analyst or data scientist and the amount of resources and time to conduct the analysis.

Step 6: communicate intelligence results

The sixth step of the HR analytics cycle involves communicating intelligence results. Real HR analytics capabilities place more effort and emphasis on telling a story about the data and visualizing the data in the context of the organization's most pressing problems and successes (Minbaeva, 2018). Storytelling can be a powerful approach in communicating data-driven insights, both in words and visually, because it stimulates emotions (i.e. the brain processes emotions differently than facts and data) (Waters et al., 2018; Welbourne, 2015). Storytelling, however, Should not be a guise or pretext for telling executives what they want to hear or “cherry-picking” the data and insights. For example, Rasmussen and Ulrich state that “HR analytics can be misused to maintain the status quo and drive a certain agenda (i.e. when you know what story you want to tell, and you then go look for data to support same” (2015, p. 237). Moreover, we need to consider the “veracity of the story” and the ethicality of the way in which data-driven insights are derived, communicated and used (Rotolo and Church, 2015). The dissemination of inaccurate or misleading insights will invariably lead to bad workforce decisions and big organizational consequences (Church and Dutta, 2013). Hence, communicating and reporting HR analytical insights involves not only some ethical interpretation on the part of HR analytics team, but also speaking truth to power.

Considerable advances have been made in linking HR to the business strategy and organizational performance (i.e. a value creation framework) through the use of the balanced scorecard (Kaplan and Norton, 1992), HR scorecard (Becker et al., 2001) and the workforce scorecard (Huselid et al., 2005). However, as alluded to earlier, not all forms of internal and external data and information can be expressed in terms of a simple metric or indicator. For example, some HR research and analytics activities (e.g. advanced research in the context of social, behavioral and organizational sciences) require advanced statistical analysis, meta-analytic, multivariant or causal modeling procedures as well as expert interpretation and insight. Nonetheless, scorecards and dashboards are effective means to communicate strategic metrics and are best used as communication and reporting tools for strategy execution once an agreed-upon strategy is in place.

Step 7: enable strategy and decision-making

The final step of the HR analytics cycle is to enable HR strategy creation and evidence-based decision-making. We have heard the proverbial mantra that behind every successful organization is a strategy that works. But what exactly is strategy? Strategy is a multidimensional concept that can be defined in many ways. Mintzberg describes strategy is terms of a plan, ploy, pattern, position and perspective. As a plan, strategy relates to leaders establishing the overall direction for the organization. As a ploy, strategy is all about maneuvering and outwitting competitors. As a pattern, strategy involves engaging in specific behaviors and consistent action to effectively implement the strategy. Strategy is also a position in terms of how an organization differentiates itself in the competitive marketplace. Lastly, strategy is a perspective that reflects an organization's culture and character (i.e. does the organization see itself an imitator, improver or innovator when it comes to its people, products and services) (Mintzberg et al., 2005). In summary, strategy involves asking intelligent questions, identifying strengths, weaknesses, opportunities and threats (SWOT), knowing the right things at the right time, scenario planning, evidence-based decision-making, establishing priorities and goals and effectively managing execution. Despite these various methods and processes, strategy in the context of HR “refers to the processes, decisions, and choices organizations make regarding how they manage their people” (Cascio and Boudreau, 2014, p. 79). A human resource strategy tends to focus on aligning people policies, practices and processes with the overall business strategy in order to achieve the organization's goals and objectives. In a perfect world, HR strategy creation should be done in concert with the overall business strategy, although this rarely occurs in practice.

A human resource strategy also involves making smarter HR decisions, and there are no shortage of models, frameworks and guidelines on the topic of HR decision-making. For example, Boudreau and Ramstad (2007) introduced their HR decision science approach to improve human capital decisions in organizations. Similarly, evidence-based HR involves making HR decisions through the conscientious, explicit and judicious use of the best available evidence from multiple sources of information (Briner and Barends, 2016). Another intriguing approach is Dulebohn and Johnson's (2013) classification framework for HR decision-making in the context of HR information systems and analytics that is based on the seminal work of Gorry and Scott Morton (1971). The framework is organized by two dimensions: (1) management decision-making level (i.e. strategic planning, management control and operational control) and (2) decision-making structure (i.e. structured, semistructured and unstructured). Their framework describes the data needs, decision characteristics and concomitant metrics across the operational, managerial and strategic levels. Specifically, this framework is a useful decision support tool for categorizing a wide range of HR practices, activities and choices in a meaningful way (Dulebohn and Johnson, 2013).

In short, these models and frameworks represent a complementary approach for enabling HR strategy and decision-making in the context of HR analytics. In our view, the primary purpose of HR analytics is to enable HR strategy and decision-making. The resultant data and insights derived from HR analytics play a central role in influencing HR strategy, decision-making and the strategic choices organizational leaders make when it comes to adopting evidence-based HR practices (Falletta, 2014). Further, we introduce the HR strategy axis as a useful framework to guide strategic HR choices with respect to human capital decision-making and investments in organizations.

Is HR analytics a new, revolutionary idea or evolutionary capability?

Arguably, HR analytics is nothing new. Yes, the data are bigger, and the technology, tools and toys are much more sophisticated and abundant. However, the use of data to understand organizational phenomena and inform HR strategy and decision-making is nothing new. In terms of historical roots, there are numerous HR research and measurement practices that have led to the emergence of HR analytics; for example:

1. Over 100 years of workforce and HR research (e.g. Schmitt and Klimoski, 1991 – a brief history of research on people at work)

2. Action research (Argyris et al., 1985; Lewin, 1946)

3. Assessment centers (e.g. Bray and Grant, 1966)

4. Data-driven methods for change (e.g. Nadler, 1977; Waclawski and Church, 2002)

5. Employee and organizational surveys (e.g. Gallup, 1988; Kraut, 1996)

6. Evaluation/ROI (e.g. Edwards et al., 2003; Kirkpatrick, 1998; Phillips, 1997; Russ-Eft and Preskill, 2009)

7. Evidence-based practice (e.g. Pfeffer and Sutton, 2006; Rousseau, 2006; Rynes et al., 2002)

8. HR decision science (e.g. Boudreau and Ramstad, 2007)

9. HR benchmarking (e.g. Glanz and Dailey, 1992; Fitz-enz, 1992)

10. Human capital measurement and metrics (e.g. Fitz-enz, 1984)

11. Personnel/employee/talent selection (e.g. Schmidt and Hunter, 1998)

12. Scorecards (e.g. Becker et al., 2001; Huselid et al., 2005; Kaplan and Norton, 1992)

13. Workforce forecasting and analysis (e.g. Bryant et al., 1973; Khoong, 1996)

These same HR research and measurement practices are widely used currently and will likely continue to serve as a significant data source in the future in terms of HR analytics work. Moreover, we shouldn't forget the social, behavioral and organizational scientists who paved the way long before the “Big Data” and “analytics” monikers, data visualization tools and data science revolution came into vogue. What is new are the availability of new technological and analytical resources and the renewed interest in analytics by organizational leaders looking for novel ways to leverage such data sources to improve HR's impact on organizational effectiveness and other important business outcomes (Putka and Oswald, 2016). This renewed interest has created a cottage industry for HR analytics products and services (e.g. SaaS-based platforms, data aggregation and visualization, apps, chat-bots, AI, deep or machine learning). Therefore, HR analytics should be characterized as an evolutionary rather than revolutionary capability with a rich history and promising future.

What is the best way to get started with HR analytics?

In terms of getting started and building HR analytical capabilities, organizations tend to be enticed into procuring R, Python and the latest HCM technology platforms with novel data visualization tools and subsequently hire a low-level HR analyst to operate these tools without clarity about their HR analytics vision, strategy and the capabilities and outcomes they hope to achieve. Hence, we suggest that organizations bring in a well-qualified HR analytics leader with the right disciplinary background and skill set before investing in any technological or analytical resources. A highly trained HR analytics leader with a strategic HR perspective can pave the way by establishing the overall HR analytics vision, strategy and capabilities for success.

Who should perform HR analytics?

HR analytics is a team sport (Green, 2017; Levenson, 2015). HR analytics practitioners and scholars come from a variety of disciplines (e.g. business management, HR and organizational behavior, industrial and organizational psychology, HR development, economics, finance, statistics, data science, computer science). While a human capital technologist, econometrician, mathematician, statistician, data scientist or financial analyst might possess the technological and statistical skills to mine, model and visualize data, we believe it takes an applied researcher with a background in the social, behavioral and organizational sciences to accurately and ethically interpret the insights derived from HR analytics in the context of individual, group and organizational behavior (Falletta, 2014; King et al., 2016).

Where should HR analytics reside?

Among larger Fortune 1,000 firms in the United States, research estimates that over 75% of these organizations have an individual or group dedicated to HR research and analytics. Nearly a third of these dedicated HR research and analytics groups report directly to the chief HR officer, suggesting that HR analytics capabilities are strategically positioned in terms of organizational structure (Falletta, 2014). Some argue that HR analytics should be situated in an enterprise-wide business intelligence function and report to line management outside of the HR function (Rasmussen and Ulrich, 2015; Ulrich and Dulebohn, 2015). We contend that this is a battle the HR profession cannot afford to lose and as such, HR analytics should reside squarely in the HR function and report directly to the chief HR officer. We also argue that the HR analytics function be led by someone trained in the social, behavioral and organizational sciences (e.g. strategic HRM and organizational behavior, industrial and organizational psychology) to ensure the discipline remains both evidence-based and ethical in terms of humanistic values (Church and Dutta, 2013). Human resources should collaborate with line management, business intelligence, information technology, finance and other functions – but the HR profession must be willing to lead the way in terms of its own strategic legitimacy and influence.

Should the HR analytics team be connected to HR strategy?

We recommend that HR analytics be merged with an organization's HR strategy function. In many organizations, a dedicated HR strategy function doesn't exist. Politically, some organizations prefer this arrangement in that all HR business partner groups and functional leaders have an equal role and responsibility in driving HR strategy creation and execution rather than a centralized function. Kaplan and Norton (2005) suggest that organizations should establish a dedicated Office of Strategy Management (OSM) to address the gap between strategy creation and execution. The notion of a dedicated OSM should be part of the corporate HR function, coupled with its HR analytics capabilities. Hence, the establishment of a “HR Analytics and Strategy” function would be ideal and ensure a coordinated effort between HR analytics, strategy creation and execution, human capital decisions and the adoption of evidence-based practices.

What are the ethical and privacy implications associated with HR analytics?

The erosion of workplace privacy and ethics coupled with a clandestine “quantified employee” agenda in an age surveillance capitalism might sound like Orwellian dystopian hyperbole. Notwithstanding, ethical questions have begun to arise about the potential abuses of HR analytics with respect to technological advancements (e.g. SaaS-based platforms that scrap and analyze external social media data, electronic performance monitoring and surveillance, wearable technologies, micro-expression analysis) and datafication of personal, and often trivial, characteristics, preferences and behaviors that have little relevance to job performance (e.g. Bassi, 2011; Falletta, 2014; Karim et al., 2015; Oehler and Falletta, 2015). In using such tools and insights, how much business value and competitive advantage can we hope to achieve and what are the trade-offs? While some insist that Big Data doesn't necessarily mean “Big Brother” (Young and Phillips, 2016), there have been a fair share of questionable and creepy stories that have made front page news. For example, according to a Wall Street Journal report, some organizations are using outside firms to predict the health risks of employees (Silverman, 2016). To what extent should employers use such data about their employees' health conditions, habits, prescription drug use and the like with respect to HIPPA and privacy laws? What about values and ethics? HR, like most professions, is built around norms, values and ethical principles. HR professionals are ethically responsible for promoting and fostering fairness and justice for all employees and their organizations as part of the Code of Ethics for the Society for Human Resource Management. Similarly, the American Psychological Association (APA) ethical principles and code of conduct require psychologists to abide by the general principle of “First, Do No Harm”.

In summary, building evidence-based and ethical HR analytics capabilities that foster trust and transparency is critically important to the success of any HR analytics effort. Each member of the HR analytics team plays a vital role as ethical researcher, analyst, interpreter, translator and educator (Dekas and McCune, 2015). Lastly, organizations need to think through how their HR analytical insights are derived, communicated and more importantly – used (Illingworth, 2015). A generic HR data governance, transparency and privacy policy is not enough. Organizations must also rapidly and rightly address how HR analytical insights will be used (i.e. determining the responsible usage of Big Data and HR analytical insights) as well as the situations and circumstances in which certain technologies, practices, data and insights should never be used.