People analytics in organizations is at a crossroads today. Analytics and data science are the hottest buzzwords in management since re-engineering and core competencies two decades ago. Companies are rushing to build analytic capabilities, both outward facing for customer engagement and inward facing for workforce management. There has never been a better time to be on the sales side of enterprise IT systems that offer business and workforce data solutions, sometimes integrated, other times separate.
Judging by all the activity and attention being paid to workforce analytics, you would think the profession had struck gold. The insights that are emerging from these new workforce data systems, especially when matched with leading data science skills by the people implementing them, appear to be very revolutionary, useful, and practical. Whether it’s insights about employee turnover, careers, workforce planning, and more, the blogs and marketing documents from consulting companies provide a steady stream of morsels for your dining pleasure.
Yet, as we all know, appearances can be deceiving. Instead of a gold mine, what we’re faced with today in the world of workforce data systems is much more like fools gold – or junk food. Though it draws us inexorably in with its great immediate taste and biological response via dopamine and endorphins, once the initial high of discovery wears off, we soon come back down to earth and have to look for another quick fix if we want to keep our energy going. The problem is that relying only on the workforce data that reside in our enterprise systems can be just as bad for us organizationally as eating only junk food instead of a balanced diet.
I see three large problems with – and offer potential solutions for – the current state of workforce data and analytics: too much people counting, insufficient historical data, and very little information on teams. Currently, in most organizations, I would estimate that only about 10% of what you need to know for proper workforce analytics is housed in standard workforce databases, systematically collected and stored on a regular basis. The solutions I propose here won’t get you to 100%, but they can go a long way toward shifting your current pure junk food diet to something more nutritious that can sustain better insights and increase your organizational health – in terms of business results, team effectiveness and employee engagement.
Too much people counting and not enough historical data
Workforce data systems are built around the need to count and keep track of people working for the organization right now. They are designed to provide a current snapshot of jobs that focuses on individuals and roles. Historical data is not preserved in an easy to access format. And the data that is collected systematically focuses on counting measures and jobs in isolation: how many people there are in a role; role titles without detailed job descriptions; spans of control; and so on.
As John Boudreau and Pete Ramstad noted years ago, workforce data systems have their roots in accounting which is why they are strong on current counting and cost measures, and weak on everything else related to workforce measurement and management. They have no information on motivation and engagement, nor on how well people work together, or their behaviors. Yet despite these glaring deficiencies, the data sitting in workforce systems is easy to access.
Given workforce analysts’ needs to be able to show some kind of insights, regardless of how meaningful, there is a huge gravitational pull to engage with the available data now and worry about imperfections later. Moreover, there is huge pressure from business leaders to show justification for the enormously expensive investments in those systems. These factors combined create a huge sucking sound as the entire field of talent analytics focuses way too much on analyses that can use those data.
Yet just because so much time and energy is spent manipulating and massaging the data in our workforce systems, that doesn’t change its fundamental junk-food like characteristics.
Consider the question of careers. Being able to accurately describe and promote careers is essential for an organization to clearly articulate the employee value proposition of joining and staying. Yet it is very difficult to construct accurate descriptions of career progression from standard enterprise data warehouses. Where historical information is preserved, it’s usually only kept active for one or two years prior; anything older is purged from the system either entirely (destroyed) or stored in backup media that are not easily accessible (tapes; disks; drives). The consequence in either case is that is it not possible to call up the entire work history for anyone with tenure longer than a year or two.
In some rare cases I have come across workforce analytics groups that recognize the value of preserving the historical information and have found solutions that work. The most reliable and accessible way for them to do so is to create their own year-by-year “snapshots” of the historical data for the entire organization, and store them in separate computer systems from the original data warehouse, where they can be accessed on demand easily by the workforce analytics group. This requires setting up an entirely different data warehouse structure than the corporate default, along with the personnel capable of preserving and working with it. To do this means investing a great deal of money, people and time to build and maintain the bespoke system.
Another part of the solution to the limitations of the data in workforce systems is to conduct workforce surveys, sometimes called engagement surveys. Workforce surveys enable deeper and richer measurement of people’s motivation and feelings about their jobs, careers and the work environment. Those data can increase by leaps and bounds the information available to paint much more complete pictures of the employee value proposition (why people come to work for us, and why they stay or leave), their experiences while working (mentoring, development, feedback, relationship with supervisor, etc.), and the workplace climate (team dynamics, leadership behaviors, the organizational culture, etc.). Yet those surveys are often done anonymously, and so can’t be matched with employee data from the enterprise workforce data systems.
Even when they aren’t anonymous, the data still is confidential and typically cannot be loaded directly into the workforce data system for widespread access and analysis. Only a select group of people on a workforce analytics team can access and analyze them. For most companies these data hurdles prevent effective use of employee surveys matched with workforce data systems.
In the handful of cases where the data are not anonymous, and are matched to employees’ individual records in the workforce data system, a further challenge is consistency over time in what’s measured. Previous rounds of workforce survey data often aren’t available and/or can’t be matched to individual records. Survey questions change over time. And the surveys are limited in the issues they do address, while business needs and the operating environment constantly shift. So today’s burning issues often are not adequately addressed in prior survey efforts. Are they addressed more effectively than working only with current workforce data? Of course, which is why they are valuable. Yet if contemporaneous workforce data covers only about 10% of what you need to know, adding in these other elements usually only gets you up to around 35-40%, and that’s when it’s done exceptionally well, which is the exception, not the rule.
The two solutions proposed here – preserve more historical data and add in workforce survey data – can fill some of the glaring holes, helping move your workforce database from pure junk food to something a bit more nutritious. Those changes are welcome if they can be done in a cost effective way. But even at their most cost effective and thorough, implemented alone they are not enough to create a truly balanced diet that will nourish all your workforce analytics needs. A missing key nutrient is teams data.
Very little information on teams
Team performance is the ultimate objective for organizations to succeed in executing the strategy and achieving operational goals. In order for analytics to best contribute to organizational success, the focus ultimately needs to be on team performance, not individual performance. Tracking individual performance and performance ratings is never sufficient to determine a team’s success. Team performance analytics require detailed information on not just the team’s performance but also its composition: who was on the team at different points in time, and who was responsible for which aspects of the team’s processes.
In an ideal world we would have detailed information on all teams: their goals, composition, performance against the goals, the roles of individual team members, and much more. Collecting all of that data consistently for all teams is cost prohibitive, and of necessity has to be limited to deep-dive, one-off analyses focused on diagnosing the performance of specific teams or groups of teams doing similar work (like sales teams, R&D teams, customer service teams, etc.). Certain information can and should be collected systematically, such as which teams exist, who is on them, when members join and leave, and similar team demographics which can be gathered and stored more cost effectively. Yet even this basic information is typically not collected systematically, and certainly not historically.
The lack of basic historical team demographic information makes the construction of accurate career paths very difficult. Just because someone might have been in certain roles over time, knowing the job title does not provide enough information to determine what they did: which teams they were on, how they contributed, and how the teams performed. So whatever conclusions can be reached based on the data that can be constructed on past work experience within the organization, it is spotty at best and usually has major holes that need to be shored up. More systematic collection of team participation data would go a long ways toward plugging those holes.
Those holes also create great challenges when attempting to evaluate leader or manager effectiveness. Managerial performance ratings, even when they include information such as 360 evaluations of their behaviors, are insufficient to determine how a leader actually performs, unless information on their team’s performance is included in the data. The reason for this is because team performance is never achieved through managerial behaviors alone, even when the measured behaviors are important inputs into team success (communication, goal setting, coaching, and so on). In the same way that accurate accounting of team membership and team performance is essential for understanding individual careers and a team member’s contributions to organizational success, that same information is needed for accurate modeling and measuring of managerial effectiveness and performance. In the absence of such data it is very hard to know whether the organization’s talent is contributing in meaningful ways. Who someone “is” as an employee or leader today can only really be effectively described by knowing their detailed history, a longitudinal description of both individual and team information that currently are missing in virtually all workforce data systems.
So is the glass half empty or half full – and how nutritious is the drink? Given the abysmal state of data in most workforce systems, I can’t help but rate those as pure junk food – covering about 10% of what we need to know. If leaders dedicate substantial effort toward shoring up collection and storing of historical workforce archival data AND surveys, and can establish consistent data libraries going back multiple years, that would improve things substantially. In that blissful scenario I would raise the rating to the bare minimum of a diet that could help sustain better talent decisions, as it would cover about 50% of what we need to know.
An almost-complete data warehouse would include comparable “factual” data on teams, including membership and how it changed over time; reporting relationships, including historical data following re-organizations that changed matrix supervisory relationships; and team performance metrics. Those data would be invaluable, and usually can be collected and stored through modest investments in reporting systems and changed managerial behaviors. If your organization were able to add this data view consistently going back many years, that would put you into the clear balanced diet range, as it would cover about 75% of your data needs.
The ideal data warehouse – when we achieve nirvana – would add to that factual data the teams counterpart to workforce survey data: survey-based measures of team cohesiveness, norms and behaviors. This last part is rarely measured but in fact quite doable because the survey measures needed for the task have been well defined by the research community. They are a core part of the work that I and many other action researchers do with organizations, drawing from decades of management research and practice. Yet comprehensive team-focused surveys are virtually never done across an entire organization, let alone repeatedly over time. If you were able to pull off this feat on top of the other optimal changes to your data warehouse, this would vault you into the rarified air of true superfood status for your workforce data system – and would cover about 90% of your data needs.
If you’re currently in junk food range, hovering around 10% usefulness, setting your sights on transforming to the highly nutritious end of the range in short order (2-3 years) is simply not practical. Like a fad diet whose effects quickly fade, you will be sorely disappointed at the results. On the other hand, setting more realistic goals is both doable and sustainable over the long term. Aim for 10% improvement per year, enjoy the short-term benefits of having a somewhat healthier diet, and yet don’t lose sight of the need to persevere and keep improving your data’s nutritional content every single year. Until you reach the ideal state there will always be big, long-term benefits from sticking to the program and improving the nutritional value of your workforce data systems.
