HR as a Research Function: the turning point McKinsey only half finishes

There is a finding in the new McKinsey survey that sits inconspicuously between the diagrams and yet describes the whole function: HR departments systematically overestimate how many of their people take part in training, and they equally overestimate how much those people care about development in the first place. It is worth pausing on why this is remarkable, because it is not just any department getting it wrong, but precisely the one whose sole reason for existing is to know the workforce. And the same survey records, right next to it, that a mere eleven percent of organizations set up their workforce planning on a long-term, capability-based footing, while the large remainder still counts heads for the next quarter.

McKinsey subtitles its annual HR Monitor 2026 a turning point for the people function, and the report means it. The function, the authors argue, stands at a defining moment where two paths open up: on one, complexity outgrows it and a growing share of its tasks migrates to IT; on the other, it rises into a leadership role that shapes how human and artificial workforces will work together. The paper rests on a survey of roughly 1,300 HR professionals and 5,500 employees across ten countries, written, as it happens, largely out of McKinsey's German offices, and for long stretches it reads like an appeal for HR to finally become strategic rather than administrative.

The obvious reading almost writes itself, and it is not even wrong: HR should stop administering forms and start researching the organization, roughly the way a development lab researches a product before building it. Only, a research operation that deserves the name has three things HR possesses in barely any of the surveyed firms: a lab where something can actually be measured, a metric that captures the result of the work and not merely its effort, and a person who, in the end, stands accountable for it. This is exactly where the report tips from a diagnosis into a misunderstanding: it treats as a maturity question what is in truth a problem of infrastructure.

Counting activity is not research, it is documentation

You see this most clearly in the matter of training. That HR overestimates how much training happens and how much anyone cares about it is not an attention error a better dashboard fixes, but the inevitable consequence of HR measuring the wrong thing. It counts how many courses were rolled out, how many policies introduced, how many mandatory trainings completed, because all of that counts cleanly, and it almost never measures whether, on Tuesday morning, any of it actually changed what someone does. Twenty-four percent of employees, according to the report, say they take part in no training at all, and more than half receive feedback once a year at most, and neither ever shows up in a metric that reaches the board. A function that counts activity instead of effect is not a research operation, but a compliance administration with better vocabulary.

And here is the genuinely uncomfortable thought. The whole report, data-rich as it is, rests on a silent assumption, namely that the gap between ambition and reality closes if HR simply gets better, more ambitious, more data-driven, more AI-capable. That is the language of self-improvement, and it misses the mechanics, because behavior in organizations is not a character trait of the workforce and not a maturity question of the HR department, but a system output: people do in the morning what their environment makes likely, and that environment consists of defaults, incentives, feedback signals and identity cues that were all set by someone at some point, only almost never with a view to which behavior they produce. Whoever fails to design that environment and instead exhorts the workforce to be more agile is doing the same thing as a gardener who shouts at the plant instead of changing the soil.

The layer McKinsey circles without naming it

Every organization McKinsey surveyed has an architecture for money with a CFO, one for technology with a CTO and one for law with a General Counsel. For the one domain from which the outputs of all three first arise, namely what people actually do, it has nothing of the kind, no dedicated role, no budget of its own, no review process. I have elsewhere called this structural gap the Behavioral Infrastructure Gap, and the HR Monitor is, without ever using the term, its most detailed survey to date. Because McKinsey's two paths are, on closer inspection, a single one: HR rises to a shaping force precisely when someone owns the behavioral infrastructure as a function of its own, and HR is absorbed by IT precisely when no one does.

The report ends with the advice that HR must build a fact-based view of its own maturity and translate it into clear priorities, and that sounds reasonable until you ask who actually does it. In finance, no one asks that question, because the answer has been CFO for a hundred years. For behavior that answer does not yet exist, and as long as it is missing, every upgrade of HR remains an appeal to a function handed a responsibility without the infrastructure to carry it. The post that would have to own this layer appears in almost no org chart, and that is not the oversight of individual boards but the consequence of behavior fraying as a category between HR, IT and leadership development, without ever fully belonging to any of them.

What research at the HR department actually demands

If you take the turning point McKinsey names seriously, and not as one more call for more ambition, then research at the HR department begins exactly where it begins in every other research operation, namely with a diagnosis of the current state before anyone rolls out a solution. At Engaginglab, the Behavioral Systems Analyzer makes visible which defaults make the wrong behavior the most convenient option, which incentives produce the opposite of what they promise, and where the signals are missing that would reinforce desired behavior in the first place. That is not a consulting opinion but a system diagnosis, on the same logic by which a financial auditor examines capital flows, and only after that diagnosis does the question of design arise. Whoever instead rolls out the next training program keeps confusing the completed hour with the changed action, which is the cost of measurability that almost every organization pays without ever putting a number on it.

HR will not rise or fall in 2026 because the function wants too little, but because the layer that actually matters still belongs to no one in most organizations. McKinsey described the turning point precisely and set the lever down beside it. The lever is no longer ambition, but the question of whether someone finally designs the behavioral infrastructure before the next expensive transformation fails at exactly the point where, to this day, no one is responsible.