People Analytics is in a Midlife Crisis (Part 1)

Our field was founded on a metric that has lost its meaning.

Recently, a colleague told me something I had never heard before: Their company eliminated its people analytics team. Not downsized, not restructured. Eliminated. A function that had existed for years was dissolved, the work redistributed, and then evaporated. It was, my colleague said, pretty quiet, which might be the most unsettling part.

It may not be the first time something like this has happened, but it was the first instance I had heard of. Ever since, I’ve been thinking about what this means — not just for that company, but for all of ours. People analytics spent the last half a generation growing up: raising teams, pursuing capability and credibility, and cultivating a function that hadn’t existed before. But the world we were born into is disappearing. If we fail to reckon with the changes we’re seeing, I’m afraid we may disappear too.

The two of us have been in and around people analytics for fifteen years. We’ve watched the field grow from a novelty to a necessity for any serious HR function. And yet, there are signs that people analytics is plateauing in some existential way. The field has created tremendous capability, but we anchored our legitimacy to a particular set of conditions: cheap capital, companies blitz-scaling, and — as an unintended result — employee headcount became the primary unit of analysis and strongest justification of our existence.

Those conditions, as it turns out, had a shelf life. The shifts in interest rates and macroeconomic conditions curbed headcount growth and spurred layoff cycles, and generative AI continues to bend the relationship — or at least, the perception of the relationship — between workers and productivity. In this first piece in the series, we explore how productivity and financial value are becoming untethered from employee headcount. We discuss the implications of a field that predicated its raison d'être on a metric that is losing its meaning.

Growing up in a Boom

On the surface, people analytics has never seemed more relevant. Everyone knows data is the “right” answer. Every CHRO evangelizes data-driven decision-making. Conference keynotes invoke the power of AI — and, transitively, data foundations — as the vehicle for workforce transformation. The vocabulary has fully penetrated the field: real-time insights, predictive models, and evidence-based talent decisions.

But professing belief in data and actually anchoring decisions to it are different things. According to Insight222’s People Analytics Trends research, 71% of organizations invest in democratizing people data but less than half report high adoption within HR, and — pretty staggeringly — only a quarter report high adoption outside HR.¹ This is the tenor of the field right now: People analytics is at once enthusiastically endorsed and yet vastly underutilized.

Data on trends in the market reveal a similar paradox. The people analytics tech market is growing — $10.4 billion in 2024, up from $8.1 billion the year before — but the rate of growth has decelerated sharply, and customer NPS has fallen 23 points in the last three years, according to RedThread Research State of the Market reports.² In recent years, the vendor landscape has seen more consolidation than expansion, with several prominent independent platforms absorbed into larger tech suites, where analytics becomes a feature more than a central source of value.

And if you really want to know where a field stands, look where the jobs are going. According to data from Lightcast published by Cole Napper, the number of roles in the field expanded roughly three-fold between 2007 and 2022, then plateaued and contracted nearly 10% in the years since.³ Additionally, in the past few years, a notable cohort of senior practitioners has exited corporate roles for consulting or fractional work, or left the field entirely. It’s anecdotal, but a meaningful canary in the coal mine: many of the people with the greatest optionality are choosing to leave.

It’s tempting to dismiss this as a sign of the times. Yes, the macro environment has been punishing, and HR at large has felt it. But there’s reason to believe people analytics has been disproportionately affected. Another study by Napper and Revelio Labs found people analytics positions are more vulnerable to interest rate increases than other HR functions, like payroll, and those who left people analytics for other fields saw greater wage growth than those who stayed.⁴

A sensible read of this data is that people analytics could ride the tailwind of a friendly economy, but when the trend reversed, we were confronted with harder questions about the true value our functions, our tools, and our talent were providing. And the uncomfortable answer is: perhaps less than we thought.

A Crisis of Meaning

When Genghis Khan reorganized the Mongol army in the early thirteenth century, he did it using headcount. A leader’s rank — and transitively, his power and status — was determined by how many men he commanded: ten, one hundred, one thousand, ten thousand. Throughout history and into modern organizations, the link between a leader’s team size and their power has operated in much the same way. Until, perhaps, now.

People analytics entered the world and operated with the same assumptions and, by doing this, we made two related but distinct mistakes.

Headcount as the unit of analysis

The first mistake was making headcount our primary unit of analysis. We built our capabilities around measuring the workforce as a population: attrition rates, spans of control, performance distributions, engagement scores, time to fill. These metrics inform legitimate questions, especially for fledgling companies and analytics functions, about how people flow through organizations. But they are also more reflective of operation and volume rather than value and impact.

Even the more “strategic” flagship use cases, like strategic workforce planning, are built around headcount questions: How many people will we need, in which roles, on what time horizon? We have become incredibly skilled at describing the workforce without connecting that description to what businesses care about most: what people actually do and the value that provides.

For years, the familiar people analytics maturity model — the progression from descriptive analytics to diagnostic to predictive to prescriptive — has offered the promise of becoming more strategic. But if descriptive analytics of population-level measurement are the foundation, even the “prescriptive” mecca falls short of answering the most important questions.

Headcount as the justification of existence

The second mistake was subtler but more insidious. People analytics didn’t just focus our analyses on tallying people — we tethered the justification of our existence to headcount and headcount growth. In other words, people analytics applied headcount logic (à la Genghis Khan) to itself, the implicit claim being that output and value scales with population size. In practice, this means that our function’s political currency was derived from the size and growth of headcount rather than the actual value we provide.

And it worked, for a while. As companies blitz-scaled between 2019 and 2022 during the era of cheap capital and the talent arms race, the value proposition of people analytics sold itself, and we scaled commensurately with the workforce. Any insight on attrition or team effectiveness was multiplied by the sheer size of the employee population it applied to. What felt like genuine product-market fit was at least partly an artifact of the index we hitched our value to, and how that index was performing in the market at the time.

But the post-ZIRP shift toward flat or shrinking headcount, increased hiring discipline, and the early signals on the impact of AI on the demand for knowledge workers exposed the fragility of this assumption. Suddenly, the math didn’t math.

Headcount diverges from output

The data on this is striking, and it shows up at every level of analysis, from the whole economy down to individual sectors and companies.

The Bureau of Labor Statistics (BLS) tracks output — the inflation-adjusted value of goods and services produced — and hours worked across the U.S. workforce as part of its Productivity and Costs reporting.⁵ When we index to a common baseline from the last ten years of available reporting we can see, from 2014 to 2022, the two lines moved roughly together. Output was growing slightly faster than hours worked, but the gap was narrow enough to suggest more people still meant more production. Both dipped in the fallout of COVID, though hours worked fell more sharply than output, which suggests the labor market was less robust to the impact than economic activity. Starting in 2022, both output and hours worked rebounded, but at visibly different rates: Output has climbed steeply and steadily while hours worked rose more slowly and at a decelerating pace. By 2024, the gap between the two lines had reached the widest point, and the trend shows output moving in a different direction from the time required to produce it.

We see a similar trend when we zoom in on revenue per employee in SaaS, an influential industry that acutely feels the impact of the changing cost of capital and role of AI in reshaping work. Across more than a thousand private software companies surveyed annually by SaaS Capital, median ARR per employee reached $130,000 in 2025 — up roughly 37% from 2019, and the increasing trend holds across self-funded and equity-backed companies.⁶ These patterns become self-reinforcing: Companies benchmark to new expectations, and the trend continues.

We can even find the same pattern in operating income — what a business generates after costs — and headcount at individual companies. At half a dozen of the biggest tech and enterprise software companies over the last decade, the relationship was linear and fairly tight before 2022. But after, it decouples: Headcount at all but one of these companies goes down or stays flat, while operating income increases, dramatically, in most cases.

Why is this happening? There may not be a single definitive reason, but there are a few factors worth considering.

One of the more proximal causes is that companies aggressively overhired in 2020 and 2021, and the efficiency gains in 2023 and 2024 are partly the arithmetic result of correcting a self-inflicted mistake. But if this were the only contributing factor, we wouldn’t expect the output-headcount gap to continue widening, so this points to other factors at play.

AI may be the most discussed explanation but probably the most opaque. It’s difficult to isolate the impact of AI from the overhiring trend that unfolded in parallel, and it’s hard to say if the gains are actual or just anticipated. Are companies reducing their workforces because AI has made them genuinely more productive? Are they reducing overhead to defray the costs of investing in AI? Or are they simply exploiting the AI narrative to tighten their belts?

Here’s what matters for people analytics, regardless of which explanation turns out to be correct: The correlation between headcount and output is no longer self-evident, and that is reshaping how organizations make decisions. CEOs are pursuing efficiency over growth. Boards and Wall Street no longer view headcount expansion as evidence of market expansion, and they often reward headcount reduction instead. Whether the underlying economics will support a permanently leaner workforce, organizations are behaving as if they do.

For a function built entirely around analyzing and managing a headcount-scaled set of problems and value propositions, this is not some methodological or peripheral development. This is an existential threat. If headcount is no longer a reliable proxy for organizational capability, if a company can shed a fifth of its workforce and double its operating income or hold headcount flat while output accelerates for three consecutive years, then the questions people analytics was built to answer are not the most important questions anymore.

Born Under a Bad Sign

None of this happened by accident. For most of its existence, people analytics has been set up in ways that made the current moment predictable, if not inevitable.

The product-market misfit

The investment decision often sounds something like: We’re scaling fast, we need to be more data-driven, let’s hire someone who knows how to do people analytics. It’s a reasonable instinct but it’s not a strategy, and the result is a capability in search of a use case.

Founding people analytics leaders are like entrepreneurs: We often arrive with ill-defined goals, opaque customer expectations, and minimal resources, into a market that has already built its own workarounds. Finance has its own workforce models. HRBPs have their spreadsheets. Executives have their instincts, which have been rewarded by the success of growing a company large enough to warrant a people analytics function in the first place.

When startups face product-market fit problems, the playbook offers two moves: stay small and iterate, or invest in customer education. Neither has worked well for us.

The “stay small and iterate” path assumes a runway that corporate functions don’t have. Startups can survive on conviction and a small burn rate while they search for fit. But businesses that hired a new team expect returns within months, not years. A team of two or three analysts running experiments and building credibility rarely survives its first annual planning cycle if it can’t point to something concrete: reports and dashboards.

And in practice, many CHROs had every incentive to support people analytics as a reporting and dashboard factory — visibly useful, administratively tidy, and safely non-threatening. A team that tells the CHRO their strategy isn’t working is considerably more valuable than one that produces polished decks for business reviews. It is also considerably more dangerous. So the patient, challenging iteration that might have led to meaningful impact rarely survived that dynamic. Most teams skipped it entirely and jumped straight to building things that looked like outputs — because that’s what the organizational structure rewarded.

The “invest in customer education” path failed more subtly. Executives don’t lack belief in data. They lack belief that people analytics has anything to do with their actual problems. If you’re a business leader wondering whether your compensation philosophy is costing you talent, you call Total Rewards. If you’re worried about org design, you call a consultant. You do not think to call people analytics — even though we see across the entire employee lifecycle and are arguably best positioned to connect those dots. Much of people analytics never solved that branding problem. We remained the team that looked backward, that was useful when you already knew what you were looking for, invisible when you didn’t.

The vendor ecosystem made this trap harder to escape. Most major people analytics platforms were built around the same imperfect unit of analysis: employee headcount. Their product roadmaps became our internal roadmaps; their definition of analytical sophistication became ours. And when those definitions are identical across the entire market, no one thinks to question them — especially when counting people is much easier than measuring their impact.

The identity crisis

There’s another factor, and it’s uncomfortable to talk about, which probably means it’s important. Most of us chose this field because we care about people. We arrived with a facility with data, grounding in psychological principles, and the ability to frame abstract questions as testable hypotheses, and we used those skills in service of understanding and supporting the employee experience and dignity.

That is not a naïve or embarrassing thing. But often it meant, subtly and over time, that we measured what we cared about rather than measuring what the business cared about. I spent years in my early career pursuing my own analytical agenda — the questions I found interesting and important — rather than relentlessly chasing the questions that might have changed the most important decisions being made around me. My arrogance ran deep and was largely invisible to me at the time. I suspect I am not alone in this.

None of this is to say we shouldn’t measure headcount (we must), or that we need to abandon our humanist principles (we shouldn’t). It reminds me of the way we refer to passengers on ships or aircraft: How many souls on board?

Among the professionals who devote our careers to understanding how organizations move — finance, ops, and others — people analysts are uniquely oriented toward understanding the work in human terms. It’s right there in our name. But we must challenge ourselves to move past using the human unit as a self-evident proxy for productivity and business value. Organizations don’t ask: How many employees do we have, where do they sit in the org, and are they engaged? but rather: Which of our investments in talent are actually creating value, and how do we know? Our field has been much less well-positioned to respond to the latter.

Our Second Act

Whenever we’re confronted with a strategic challenge, it seems the instinct in people analytics is to reach for an operational solution. More tools. Better models. Faster time to insight. AI is particularly tempting on this front — the catnip is real, and the field’s operational orientation makes it easy to believe the next technology revolution will solve what the last one didn’t.

It won’t.

By technical measures, people analytics is more capable now than ever before. But we cannot grow our impact if our legitimacy is measured in a currency that is dropping in value. And although some economic trends are cyclical, the expectations about the relationship between headcount and output are unlikely to revert. Instead, we’ll likely see revenue per employee continue to rise, and maybe it will cease to become a meaningful metric altogether.

The solution will not come from a new tool. We must answer reframe the questions we ask, the outcomes we anchor to, the decisions we influence, and the very identity with which we approach our work.

Part 2 will be about what that shift looks like in practice, and what it will mean for people analytics to create a second act that is more honest, more useful, and less vulnerable to vicissitudes than the first.

---

This post was coauthored by Colby Kennedy Nesbitt and Yuyan Sun. You can find more of Colby’s work at Variance, Explained and more of Yuyan’s work at Amazing Work!

1. Insight222, People Analytics Trends 2024 (December 2024). Summary published by Naomi Verghese, “Measuring Value and Driving Adoption of People Analytics Products,” myHRfuture, December 4, 2024. https://www.myhrfuture.com/blog/measuring-value-and-driving-adoption-of-people-analytics-products

2. RedThread Research, People Analytics Technology: State of the Market 2025 (May 2025). Press release: https://www.prnewswire.com/news-releases/redthread-study-reveals-people-analytics-tech-market-hit-10-4b-in-2024

3. Cole Napper, People Analytics: Using Data-Driven HR and Gen AI as a Business Asset (Kogan Page, 2025).

4. Cole Napper, Jin Yan, and Ben Zweig, “What Is Happening to People Analytics? A 15-Year Trend — Part One,” Directionally Correct (Substack), September 2, 2024.

5. U.S. Bureau of Labor Statistics. (2025). Productivity and Costs, Annual Release, Nonfarm Business Sector. U.S. Department of Labor. https://www.bls.gov/productivity/

6. SaaS Capital. (2025). 2025 Revenue Per Employee Benchmarks for Private SaaS Companies. https://www.saas-capital.com/blog-posts/revenue-per-employee-benchmarks-for-private-saas-companies/