Introduction
After years of doing workforce analytics, as much as we talk about it, there are still impediments to its effective adoption and integration into the business. This resistance is often deeply embedded in the psyche and structure of the Human Resource Organization as well as the larger organization's view of HR and the supporting infrastructure of the company. Before diving into what may impede workforce analytics adoption, let's define what it could be. Early in my career, I was fortunate enough to work with a chief human resource officer who described the work of human resources as turning the company's strategy into the workforce's behavior. He understood that analytics was not just a tool but a vital aspect of being a strategic partner, underscoring the importance of HR in the company's success.
Workforce analytics cuts across the whole organization. It is the ongoing and active investigation of how the company: 1
- Organizes their workers to use technology to be efficient, effective, and profitable as they compete for resources and markets.
- Builds and maintains an inventory of workers with the knowledge, ability, tools, and interest in doing the jobs that make up that organization.
- It provides the managers with the tools to guide and motivate those workers.
Innovates to respond to technological changes, competition, and the resources they need.
Silos Within HR May Inhibit Helping Line Functions Solve Pressing Issues.
The first impediment to embedding workforce analytics in the organization may come from the structure and thinking of the HR function itself. Early in my career in workforce analytics, I consulted with a company (we shall call XYZ) undergoing a reorganization in response to competitive pressure. A series of interviews was revealed with the plant managers and business unit heads affected by this large-scale reorganization. First, someone from organizational design and effectiveness would ask questions about what they do and need, followed by employment and staffing, compensation, equal employment opportunity, and industrial relations. The heads of these plants all said the same thing; each representative asked similar questions but in very different languages, which left these leaders confused and frustrated. They knew they needed help and wanted to act in a way that would balance the need to be more efficient while attending to the needs of their workers. They could not understand why they could not deal with one person who had one view of how HR could give them fact-based insights to understand their options. While the role of the Business Partner is designed to solve this problem, the separate thinking of the HR functions, the disparate systems, and the data that support them make an integrated approach difficult.
How Some Human Resource Functions See Their Role
Some HR functions have a more limited view of workforce analytics than XYZ. They avoid dealing with line functions. Recently, I led a meeting of people tasked with running the workforce analytics function in many large national companies. We invited a speaker to talk about the analytical work done to help improve the productivity of a large unit of a global company. This presentation demonstrated the unified approach of the HR function. Our speaker talked about adopting new technology, changing roles to gain the maximum efficiency from that technology, pushing down decision-making to the front-line workers, realigning, retooling, or laying off workers who could not or would not adapt to the new role, rethinking the role of leaders, all of which in the end led to a happier and more productive workforce. Using the results of their analysis, they mapped how to get from where they were to where they needed to be.
Many of the attendees' reactions were surprising. Many did not think that what the speaker presented was a matter HR should handle. They said the speaker was dealing with an operations problem, and once the operations folks decided how they would organize, HR's job would begin. In other words, HR would react to operations and not collaborate with them and take on the responsibility of working to guide the business.
The Inertia of HR's Traditional Role Led to Silos
To some extent, this is not surprising. HR has been tasked with keeping the company out of trouble by greasing the wheels of a system already in operation. This preventative effort involves complying with laws and regulations, balancing the cost of the workforce with the need to find and keep good workers, and ensuring managers respect their workers and use the rewards systems fairly and effectively. Finally, it involves supplying information and services to the workers: payroll, benefits, training, etc., and how and when they need it. A whole set of specialized professional organizations 2, academic institutions, and journals support each HR function engaged in these activities. Indeed, designing and evaluating compensation programs, employment and staffing processes, and approaches to training and development are meaty topics. At the same time, this self-examination contributes to siloed thinking. It supports best-of-breed systems that structure data specific to the function's needs and challenge data integration. This division of labor also segments the HR budget to purchase products and services that are affordable and uniquely tailored to the individual HR functions. It inhibits collaboration and a comprehensive view of the HR function.
The Role of HR as a Worker Advocate
HR often needs help combining their role as employee advocates with the scientific approach to the study of work. Taylorism, or what in modern parlance is seen as the maniacal scientific focus on labor force productivity and efficiency, is seen as antithetical to the interests of workers. Contemporary press reports from warehouse workers 3 and customer service workers 4 in call centers reinforce this view of analytics. I remember hearing one modern critic say that if Charles Dickens had written today, he would have set his stories in call centers and warehouses. This fear of analytics is ironic. Modern workforce analytics practice grew out of employment law, the uniform selection guidelines on employee selection, and the need to ensure fair treatment of all workers. Its goal was to eliminate bias that led to disparate treatment and short-sighted business decisions whose impact was detrimental to the workers subject to them.
Indeed, the thoughtless uses of data and productivity measures masquerading as analysis have resulted in unintended and negative consequences for many workers. A recent example comes from call center operations. In one large call center, it was found that a small proportion of the CSRs outperformed their coworkers by a wide margin. These measures included successful call completion, completion time, satisfaction ratings, percentage of disconnects, and add-on sales. It was decided that these superior CSRs would set the performance benchmark for all other CSRs. After all, everyone was hired to meet the position's basic requirements and given the same training, so why would some CSRs perform so much better? It was felt that those who did not meet the standard would have to work harder. The problem was that no matter how hard these folks worked, they could not match the productivity of the top workers. They were feeling oppressed by a standard they could not meet. In this case, the CSR performance being measured did not reflect what was being done and was detrimental to customer service.
The company skipped an essential step in all workforce analytics by not asking why those CSRs were able to do so well. Once they asked that question, they discovered that these folks had figured out how to navigate and efficiently link the diverse systems and information stores supporting their work with customers. These learnings could then be translated into better training and systems integration. This not only improved performance but also the happiness and engagement of the workers.
The Challenge of Getting and Integrating Data
Data integration and quality is a challenge. Professionals doing workforce analytics face many of the same issues that have confronted building reporting systems and business intelligence applications for years. The number of separate HR systems that support the function can be staggering. The need to feed data to supporting applications can cause problems when data is reintegrated into more extensive databases later. Unique keys to integrate data from diverse sources often do not exist or are lost or changed in the process. Adding additional data from Finance, Sales, Marketing, and Operations can be more challenging. Arguments about data quality are not often about errors in the data as much as definitions and the use of the data. For instance, integrating data from HR and Finance can involve discovering that finance thinks of and accounts for employees differently from HR. Arguments about how many workers are in a department or division often come down to how finance sets up the general ledger for accounting versus how HR views the reporting relationships of workers in that organization. Unfortunately, those discrepancies can be heated and lead to a distrust of the results of workforce analytics. Similar confusion may come from the unique needs of systems that separately support operation – scheduling or sales management – rapid adjustment of territories and those assigned to them or tracking the composition of sales teams.
But Don't Business Intelligence Systems Do All Of This?
Yes, but these systems and the infrastructure that support them often need clarification with the more inclusive category of workforce analytics. First, they are usually housed in business intelligence departments, which are meant to house business analytics, and these systems have limited analytic capabilities. Scorecards, dashboards, and the data warehouses that support them often integrate HR with Finance, Sales, and Operations data. These systems solve many challenges of extracting, validating, cleaning, transforming, and distributing valuable information. They allow executives and managers visibility into whether they are achieving their goals in the form of KPIs and provide the ability to see how parts of the organization, geographies, and time frames contribute or fail to deliver on those KPIs. They often contain human resource information: workforce demographics, education, training, experience, tenure, etc. However, they frequently fall short of providing insight into why things are going well or failing and showing what are the best ways to act to improve performance.
They are control systems designed to show management what is working and where it may need to be fixed based on the business model they instantiate. They are sophisticated reporting systems that grew out of accounting principles associated with sales, finance, and operations, which represent how managers think about their work and are held accountable. They are often top-down in their logic, decomposing the components of profitability, CRM, Operations/Supply chain, and the workforce. They are necessarily supported by a complex assembly line process that often includes daily extraction, validation, cleaning, integration, and data transformation. This must be stable and sustainable over years of service, securing information distributed to a broad audience. The demands for building and maintaining these systems require skills appropriate to building complex infrastructure and thinking more attuned to spreadsheets than statistical or mathematical modeling. 5 The product of workforce analytics is different. It requires a different set of skills. Its product is a set of unique, quickly-produced mathematical models that answer a specific question. These models will need interpretation and require thinking through how the business can best act on the implication of that model. This differs from an easily understood graphical user interface required for a broader audience.
Workforce Analytics Faces Its Problems in Explaining What It Offers the Corporation.
First, what is the product? Data warehouses, scorecards, dashboards, and the infrastructure that supports them are tangible. Workforce analytics is a process that leads to fact-based models and knowledge about how to improve the business. These models can provide immense benefits but are conceptual, requiring interpretation and concrete action. While workforce analytics can be packaged a bit, it is still a deliberate process that includes the following:6
- Understanding the problem to be solved, the background that may have led to the problem, and the business context for the solution.
- Identifying the workforce and business factors likely to explain what is happening and the possible solutions.
- Capturing, cleaning, integrating, and relevant data. This process is similar to that described for scorecards and dashboards but needs to be quicker and more flexible. Presentation and distribution are also much simpler because the audience is usually limited to a small group of stakeholders, requiring minimal data security.
- Applying analytical methods to build and validate models that explain what is happening and offer options to improve performance. Stakeholders tend to think of analytics as a monolith. However, there are many approaches to workforce analytics. Analysts must carefully explain how their methods are appropriate to the problem being solved. The approaches vary. 7
- Industrial Psychology and Organizational Development—Building a more efficient and effective organization by reorganizing around new technologies and understanding their impact on jobs, capabilities, and compensation. These methods may include surveys, factor analysis to summarize capabilities, various forms of clustering to group people into employment, and running scenarios to test operations.
- Optimization Algorithms—These solve the inevitable problems of labor force supply and demand by balancing the timing, costs, benefits, and risks associated with retaining people with the needed skills, developing current workers to acquire the skills, hiring those with the required skills, or using temporary, contract, or outsourced labor to fill the gaps.
- Process models—These are associated with understanding how workers move through the organization, detecting barriers to movement, and improving the internal labor market.
- Various forms of regression and modeling are associated with understanding who makes a good worker or the effects of training and development.
- Network analysis /relational analytics – The modeling of communication patterns to map the behavior of teams and the value of the people in them8
- Various forms of Machine Learning—to better automate routine activities like navigating the complex systems and databases that customer service agents contend with and detect innovation naturally occurring in the organization.
- Interpreting and presenting the results of those models and the options to improve performance contained in those models to stakeholders to get their feedback and buy-in.
- Finally, implementing and evaluating the changes to policies, procedures, technology, and management actions required to achieve enhanced performance, which may include scorecards or dashboards.
Accomplishing all of this requires skilled and experienced analysts who can bridge the gap between the abstract world of statistics and mathematics on the one hand and the pressures on businesses to make money, control risk, and act quickly on the other. It also requires stakeholders who value what can be learned from this effort and eventually own the decisions and actions resulting from using this knowledge. This often necessitates overcoming deeply held beliefs of executives who have successfully relied on their intuition and judgment.
The new world of Artificial Intelligence and Machine Learning will increase the number of ways to model problems and suggest ways to improve the company's and its workers' performance. Hopefully, these algorithms will make machines more intelligent and make the managers and workers who use them wiser and happier.
