Human beings’ natural instincts are a terrible source of guidance when seeking to make rational decisions. Common mortals are subject to a long list of debilitating cognitive biases, many of which are prosaically albeit neatly summarized here – worth a look.
In their 1974 report entitled ‘Judgment under Uncertainty: Heuristics and Biases’, Daniel Kahneman and Amos Tversky, who are the subjects of Michael Lewis’ latest book ‘The Undoing Project’, provide a structured review of biases affecting decision making. They organize these pitfalls of the mind into three main categories, namely representativeness (simply put, over-reliance on stereotypes), availability (over-reliance on the immediately available data which comes to mind) and anchoring (over-reliance upon the initial piece of information offered).
In the early 2000’s, the same Michael Lewis described in ‘Moneyball’ how statistical methods outperform experts subject to these biases when it comes to evaluating baseball player talent and making hiring decisions. Specifically, experts tended to overvalue athleticism, thereby creating some inefficiency in the market for players which could be arbitraged by smarter teams.
Fifteen years on, this intellectual approach to people management is taking a new dimension thanks to big data. In ‘Competing on Talent Analytics’ published by the Harvard Business Review, the authors encourage corporate officers to favor ‘people analytics’ over gut instincts to analyze workforce parameters. The report features tools designed (i) to assess the evolution of the health of an organization by monitoring employees’ engagement, (ii) to identify business units requiring management attention, (iii) to pull the most effective productivity levers, (iv) to forecast workforce needs in cyclical industries and (v) to evaluate talent. Google is often cited as a pioneer in this field with its People and Innovation Lab (see in particular Google’s Project Oxygen).
And yet, resistance to data is pervasive across industries, with many claiming with solemnity that ‘a firm’s management cannot be led by data’ as if companies were about to be taken over by an invisible and uncontrollable force. But if data analytics can bring benefit to the operations of industrial machinery why couldn’t it do the same to the management of human resources? Predictive talent models are conceptually equivalent to predictive maintenance models. It is about going from intelligent machines, intelligent plants or intelligent grids to… intelligent people decisions. So what is the big deal?
At an extreme, ‘algorithmic management’ invalidates the need for certain managerial functions through artificial intelligence. Analytics may not only replace factory workers. They may replace the decision-makers themselves, in HR and beyond. Thus, for certain managerial functions to rely on data is perceived to be an auto-destructive proposition.
Another deeply ingrained notion is that objectivizing people is considered to be a crime in the collective psyche. It suffices to observe the epidemic reaction of an audience when pronouncing in the same sentence ‘data’ and ‘people’. Individuals’ urge to share personal information on social networks only occurs under the naïve cover of presumed privacy. Instinctively, data and data analytics are not only seen as a threat to employment, but also as a threat to privacy and thus to freedom.
The noble emphasis on experience, heuristics and instinct versus data must be put in perspective. Finding the ability to tap people analytics where it is safe to do so to gain some productivity and build a competitive edge must be the way to go. To take a concrete example: A McKinsey paper reports that an automated approach to resume screening would increase gender diversity towards a fairer world as gender biases are neutralized. And since gender diversity is the source of a competitive edge (see ‘Smart Teams Level The Playing Field’, March 2017), what is not to like?