The Promise That Built People Analytics
People Analytics was built on a compelling promise: better decisions through better data. If organizations could measure hiring quality, engagement, learning, performance, retention, or wellbeing with increasing precision, they could allocate resources more effectively and create better workplaces. In many respects, this promise has been fulfilled. Human resources has become significantly more evidence-based than it was even a decade ago.
Yet beneath this progress lies an uncomfortable paradox.
The more organizations rely on a metric to guide decisions, reward behavior, or evaluate success, the less that metric reflects the phenomenon it was originally designed to measure.
This is not simply a flaw in measurement. It is a fundamental property of human systems.
Economist Charles Goodhart observed this phenomenon while studying monetary policy in the 1970s. His original insight was deceptively simple: "Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes" (Goodhart, 1975). Although his work focused on economics, the principle extends remarkably well to organizational life.
The version most people recognize today comes from anthropologist Marilyn Strathern, who famously wrote, "When a measure becomes a target, it ceases to be a good measure" (Strathern, 1997).
Few statements capture the modern reality of People Analytics more accurately.
Metrics are not passive observations. They change the very behaviors they intend to capture.
When Measurement Becomes Intervention
Consider employee engagement.
Organizations frequently interpret engagement scores as an indicator of organizational health. Initially, these surveys may reveal meaningful differences in leadership quality, communication, trust, or psychological safety. However, once engagement scores become executive KPIs or influence managerial evaluations, the nature of the measurement changes.
Managers begin reminding employees to complete surveys. Survey timing becomes strategically planned. Local initiatives are introduced immediately before measurement periods. Communication campaigns intensify. In some cases, subtle social pressure emerges to provide positive responses.
The dashboard improves.
Whether the underlying organizational climate has improved is a different question altogether.
The metric has shifted from observation to intervention. The same pattern appears across nearly every major HR indicator.
Time-to-hire was introduced to monitor recruitment efficiency. Once recruiters are rewarded for reducing hiring time, the pressure to close vacancies quickly may outweigh the incentive to identify the strongest long-term candidates. Learning metrics often reward course completion rather than capability acquisition. Performance systems encourage employees to maximize measured outputs, even when those outputs only partially represent meaningful contribution. Diversity targets can unintentionally prioritize numerical representation over genuine inclusion. Wellbeing initiatives risk becoming successful because participation rates increase, not because employee wellbeing improves.
None of these metrics are inherently flawed.
The Proxy Problem
The problem begins when organizations mistake a proxy for the phenomenon itself.
This distinction is more important than it first appears.
In social science, many organizational constructs are inherently latent. Trust cannot be observed directly. Leadership quality cannot be directly measured. Employee potential, collaboration, motivation, resilience, and organizational culture all exist as theoretical constructs rather than objective quantities. We therefore rely on indicators that approximate these concepts.
An engagement score is not engagement. A performance rating is not performance. A learning completion rate is not learning.
They are proxies.
Useful proxies, certainly, but proxies nonetheless.
The difficulty arises when organizations forget the difference.
Campbell's Law and the Corruption of Metrics
Donald Campbell described this dynamic decades ago in what later became known as Campbell's Law: "The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort the social processes it is intended to monitor" (Campbell, 1976).
People Analytics sits precisely at the intersection Campbell warned about.
Organizations increasingly use quantitative indicators to allocate bonuses, evaluate leaders, prioritize investments, and shape strategic decisions. As the consequences attached to these indicators increase, so too does the incentive to optimize the numbers rather than the underlying reality.
Measurement begins to influence behavior.
Behavior begins to influence measurement.
Eventually, distinguishing between genuine organizational improvement and statistical improvement becomes surprisingly difficult.
Rewarding the Wrong Things
Management scholar Steven Kerr illustrated the same problem from another perspective in his classic paper On the Folly of Rewarding A, While Hoping for B (1975). Organizations frequently claim to value collaboration while rewarding individual competition. They emphasize innovation while promoting risk avoidance. They celebrate learning while incentivizing short-term productivity.
The mismatch is rarely intentional.
It emerges because what is easiest to measure gradually replaces what is most important to achieve.
More Data Does Not Mean More Understanding
This observation should make everyone working in People Analytics slightly uncomfortable.
The field often celebrates increasingly sophisticated predictive models, richer dashboards, and more comprehensive data integration. Machine learning models now estimate turnover risk, predict internal mobility, identify high performers, and recommend learning interventions with remarkable statistical accuracy.
But predictive accuracy is not synonymous with organizational understanding.
An algorithm may accurately predict which employees are most likely to resign without explaining why they resign.
A model may identify employees with high promotion potential while reinforcing historical biases embedded within previous promotion decisions.
A dashboard may perfectly visualize indicators that have already drifted away from the reality they were intended to represent.
Sophisticated analytics cannot compensate for weak measurement assumptions.
In fact, greater analytical sophistication may create greater confidence in conclusions that deserve more skepticism.
This is perhaps the central paradox of modern People Analytics.
As our ability to analyze data improves, the quality of our decisions depends less on computational power and more on asking whether we are measuring the right thing in the first place.
The Illusion of Objectivity
John Z. Muller argued in The Tyranny of Metrics (2018) that excessive faith in measurement often creates an illusion of objectivity. Numbers feel neutral. Dashboards appear scientific. Rankings suggest precision.
Yet organizational reality remains stubbornly complex. Not everything that matters can be counted.
Not everything that can be counted necessarily matters.
Before You Build Another Dashboard
This does not imply that organizations should abandon measurement.
Quite the opposite.
Good measurement remains indispensable. Evidence-based management is unquestionably superior to intuition alone.
However, good measurement also requires intellectual humility.
Metrics should inform judgment rather than replace it.
Indicators should stimulate questions rather than prematurely conclude answers.
Most importantly, every People Analytics professional should occasionally ask a deceptively simple question:
"If people started optimizing this metric tomorrow, would it still measure what we think it measures?"
That question may be more valuable than any dashboard we will ever build.
References
Campbell, D. T. (1976). Assessing the Impact of Planned Social Change. The Public Affairs Center, Dartmouth College.
Goodhart, C. A. E. (1975). Problems of Monetary Management: The U.K. Experience. In Papers in Monetary Economics (Vol. 1). Reserve Bank of Australia.
Kerr, S. (1975). On the Folly of Rewarding A, While Hoping for B. Academy of Management Journal, 18(4), 769-783.
Muller, J. Z. (2018). The Tyranny of Metrics. Princeton University Press.
Strathern, M. (1997). "'Improving Ratings': Audit in the British University System." European Review, 5(3), 305-321.