More and more solutions are entering the market, claiming to have developed a ‘crystal ball 2.0’ using Big Data, which has become the catch-all term. Numerous entrepreneurs are also boasting about ‘doing Big Data’.
One can do pancakes, one can even do stupid things, but one does not DO Big Data.
Behind this term is the concept of gathering large volumes of a wide variety of data, quickly, and often in real time. Yet we are quick to forget about the veracity and value of this data, often because it is difficult to evaluate these fundamental parameters. What can we draw from a range of erroneous information, the quality of which was not the main concern? Statistically, we can draw lots of things! But when it involves redesigning the Human Resources decision-making mechanism, using all this data, issues of ethics and moral codes should be considered. Along with everything else!
The second era of the Internet
Big Data is now clearly everywhere. In retrospect, we might call this the second era of the Internet. For more than 25 years, we have spent our time populating the vast ocean that is the Internet with billions and billions of data. Today, more than 29,000 Gigabytes of data are published every second, amounting to over 100 billion GB since January 1! And there’s no stopping it now. After so many years, we now have so much available data that we can analyze it and get it to tell a story. Google, for example, can predict a divorce some six months in advance, or even where you will be living in a few years!
Today, these predictive experiments are being conducted simply because they are possible. In a way, this is both fascinating and terrifying, but ultimately it is just an experiment and so we shouldn’t fee particularly targeted or powerless. These specific cases raise a number of crucial questions: should we use the power of technology just because we can? At what point do responsibility and ethics come into play? What are the consequences of this statistical approach to society? In line with what we started when we set up and developed “Le Lab RH”, I am now set on waking up as many people as possible to the major issues that determine the society we want for tomorrow. Data, and how it is used, is central to this model. Get ready for a wake-up call!
“The consequences of every act are included
in the act itself” – George Orwell
Does HR Big Data have a place in our society?
No big mystery here: of course it does. Analyzing data has the potential to deliver great socio-economic value to organizations, especially in motivating talent. As a trained physicist, I love numbers and what they can reveal. But physics aside, the words of Rabelais sum up the situation perfectly: “Science without conscience is but the ruin of the soul.” I defend the idea that, with the power technology gives us come the responsibilities of anticipating the significance of the action we have taken. And while Marvel fans will identify with this apocryphal quotation, the philosophy behind it has never been truer than at this time of profound change in our society.
It is important to remember that no technology is either good or bad. It is how we use it that determines the nature of its consequences, regardless of our intentions. The best intentions in the world can have disastrous results, that are detrimental and sometimes unfixable. This is why, when data is analyzed and interpreted in the exclusive domain of HR, a number of precautions need to be taken. And it is only now that emerging players are partially starting to consider such precautions, as underlined by Professor Jean-François Gagne (University of Paris Dauphine) in the French newspaper, Les Échos.
Human Resources management is being
transformed at break-neck speed
Human Resources management is being transformed at break-neck speed. Formerly an administrative function, in charge of managing the payroll and social risk, it has now become the function responsible for attracting, developing and retaining talent, with growing responsibility as an organizational coach. This is an entirely different story. Faced with this new order, most HR managers feel somewhat unsettled, while some are looking for solutions from new tech or advice ‘gurus’. But choosing the best time to water a field is, from an ethical standpoint, very different from managing the career of a human being! Certain aspects of how society functions must not simply be mechanized without deeply reflecting on the human aspect.
HR analytics is a normal, healthy practice in companies. But as soon as it tries to predict the future based on the past, we should take time to reflect…
The HR Big Data approach still has many flaws, temporarily calling into question the reliability and relevance of its use, at least for most of the HR function’s various aspects. Because, while HR Big Data might be a godsend for HR departments looking to work more closely with strategic and financial decision-making centers, using advanced analyses, it is fundamental not to forget that behind this data are human beings, each with their own personal potential, desires, and issues. If HR Big Data is perceived as the traditional analysis of data, I say GO. But if the aim is more a desire to predict the future and trace curves and trajectories based on the past, I say STOP. A number of issues must be addressed before this!
HR Big Data is deterministic
Determinism is a concept deriving from philosophy, whereby every event is determined by clearly-established laws and causes. Applied to career management, for example, determinism would advocate that the future occupation of each person is pre-determined. The causes can be diverse: the market’s law of averages, orientation from a very young age, etc. It is a system in which your desires and motivations would be so secondary that they would never be taken into account. Some readers might find this amusing, seeing little difference from the society we live in today perhaps?
But the reasoning behind data analysis is the desire to predict the future based on the past. Let’s imagine that we use Big Data to analyze millions of career paths around the world, in an effort to define the trends for career paths. It is important to stress here that, as with any statistical process, any ‘outlier’ path (one that is far outside the norm), and therefore atypical, is usually removed from the analysis to avoid distorting the averages and other composites. You will, therefore, observe that people who perform Job X usually evolve into Job Y, and sometimes into Job Z.
The command of the old despotisms was:
“Thou shalt not.”
The command of the totalitarians was:
Our command is: “Thou art.”
– George Orwell
If you are an HR manager fully committed to such an approach, you will soon conclude that your employees who perform Job X might evolve into Job Y or Job Z. Why? Because that is what usually happens. To hell with path diversity and to hell, above all, with the wishes and desires of individuals: society’s average has spoken! This is determinism and, at a time when one of the major challenges for companies is to rebuild employee engagement and encourage their capacity to adapt, you can safely bet that for employees this will not be the most popular measure!
HR Big Data is often ineffective
I say ‘often’, and certainly not ‘always’. Here again, I am in no way questioning the huge potential of data analysis to boost productivity and improve employee wellbeing. That is obvious. However, when we want to predict the future based on the past, we need to compare two comparable situations. And in Human Resources, this is not at all the case. As explained earlier, HR managers have been seriously challenged by their evolving profession, alongside all their colleagues!
Almost half of all occupations today will cease to exist in five years time. Meanwhile, 65% of occupations that will exist have yet to be created. Continuing with the previous example, when looking to predict the best possible career developments, it is easy to understand that more than 80% of the occupations included in an analysis of past careers will be different in the future, between those that will disappear and those that will emerge. On this basis, how can any prediction be valid? The predictive approach in a field like career management is extremely restrictive and could even be harmful to the company’s future competitive strength, creating erroneous recommendations that are neither linked to performance nor the prospective changes in occupations.
In this sense, predictive analysis is often ineffective because it compares two worlds that share very few commonalities: one is at the dawn of the digital age, the other a fully digitized world. Under other circumstances, in a world that is likely to change little, the predictive approach would be extremely relevant. But since the whole framework is set to evolve, it is the same as in science when there are too many uncontrollable parameters to draw a conclusive analysis from an experiment! And any patches for the methodology that might be added later will not detract from the fact that we should know our limits when managing the life of a human being!
First and foremost, behind each of these predictions is the all-powerful promise of predicting performance based on a set of criteria for which there are measurements: presenteeism, sales revenues, etc. But, by definition, a prediction is only valid for what is accounted for. I recommend that considerable thought be given to everything that is not taken into account, either because it is not measured or is difficult to measure, but which still has a strong, long-term impact on the immaterial and sometimes material capital of the company concerned.
For example, if a selection process is used to recruit salespeople based on sales performance, is there not a risk of undermining the team’s social relationship? Or of leaving out some of the support profiles capable of making the ‘stars’ perform so well? Performance is the result of a process that is much more complex than the individual criteria observed in sales figures alone. Predictive analysis is therefore based on short-term or even medium-term performance, but certainly not on long-term performance.
“The first time this can be called a mistake,
thereafter it is pure obstinacy” – Alain Leblay
For certain strategic aspects, where the environment is changing less, HR Big Data can be very valuable. For example, it is possible to predict the likelihood of an employee resigning, to help anticipate this and thus find ways of retaining them. It is also possible to assess changes in the payroll over time based on employee profiles, by knowing the increases generally needed for maximum employee retention. And so on… For all adjustment parameters, HR Big Data is, in fact, a valuable asset. For the rest, it is better not to play the sorcerer’s apprentice. This is true, for example, of the experiment by the MIT, which recorded the conversations of employees to reveal the changes in their voices and the duration, to show variations in their level of engagement. But where does this lead us???
Existing HR Big Data does not generate engagement
This seems like quite an obvious statement, but if you knew that each element of your professional life was ruled by an algorithm, what enjoyment or excitement would you have thinking about the future? If it was all written down in advance, why bother to make an effort? The predictive analysis of data will not generate engagement over time. And even if we accepted that this was possible, what would the human and social cost be? If the aim is to generate engagement, why not simply ask those present for their opinion rather than analyze the past lives of people? We all have different objectives, affinities, and motivations, and it is healthy for the HR function to incorporate this upstream in its strategy to generate engagement and, in so doing, to develop and retain its ‘talent’ more effectively. This is common sense.
Solutions that do not use the human aspect by involving people when gathering information generate little engagement in return because they have not paid any special attention and have not, therefore, called on the reflexive and emotional centers. The emotional center, in particular, plays a key role in the process of engagement and this is why current thinking in marketing tends to contextualize the approach by involving consumers more, for example. In HR, it is more a case of presenting an experience that engages people, getting them to talk about themselves, for example.
An HR Big Data approach with an in-depth analysis of behavioral data can help to reduce some of this problem, like neuro-marketing strategies that harness this kind of data for consumption. But such behavioral data has yet to be gathered since the vast ocean of the Internet does not directly contain such information. While the psychological profiles of employees can certainly be identified by studying their lives on Google or Facebook, from an HR perspective, I will let you explain it to your employees:).
HR is neither a marketing nor a finance function
and employees are not
just customers or entries in a ledger.
Nothing beats transparency and co-creation. When we want to achieve something, we get other people onboard and explain the project to them, asking for information needed to succeed. A company that generates engagement tomorrow will likely bring more people onboard this way than if they acted like Big Brother! The HR function can be a partner in this business. But if it becomes the famous business partner described by Dave Ulrich some fifteen years ago, it will lose its social role, a role that is one of its primary missions.
And what if we started to apply matching before predictive HR?
Big Data is a strong trend whose potential is still underestimated. But for HR, some lines must not be crossed: preserve the social dimension is hugely important. And yet data can be used effectively without needing to use the predictive approach characterized by Big Data. This analytical, non-predictive approach, can be applied by matching, or in other words comparing its compatibility with a finite list of criteria. And, in my view, this is healthy and sufficient enough to effectively equip more than 80% of HR applications today and tomorrow.
The principle is simple. Just as in school, to solve an equation you must determine a certain number of parameters. Applied to our subjects, a set of information or criteria is needed to determine an outcome. For example, if you want to know if a job offer matches your profile, you will need to know if the criteria that define you are relevant and compatible with those presented in the job offer. Far from being predictive, a method that includes unknowns, your friends, your parents, and even your financial advisor to forecast your future, this is all about you!
With a reasonable amount of information about you, very advanced recommendations can be made that are not based on prediction, merely by considering compatibility on a ‘point by point’ basis. There is also another much less assuming and more pragmatic approach than Big Data, which is Small Data, as explained in this article by an expert at Manpower.
And it would, honestly, not be useful to have too much information about you because the market would not know how to use it. In a changing world, the future is very uncertain and could not be determined today based on the past. This uncertainty also limits the ability to define the parameters of tomorrow’s occupations, businesses, and professional development! So, once again, let’s keep it simple. I would be more than willing to recommend that our friends in HR invest massively in setting up a talent planning (or workforce planning) unit, to work on the skillsets and occupations that will be needed tomorrow, rather than on the career paths of tomorrow, in order to consolidate the competitive strength of their companies.
A company’s competitive strength tomorrow will depend on its capacity to attract, develop and retain its employees. Knowledge is no longer enough, engagement is key!
To conclude, I have three words in my mind: realism, diversity, and serendipity. Realism, because it is more than necessary to be realistic nowadays, and in this article, the Guardian reminds us that, however necessary they might seem, statistics do not predict the future! Then comes diversity, and as stated in the movie, “Lily Sometimes” (2009): “If you’re too eager to fit into the mold, you end up looking like a pie.” This is not a situation I would recommend.
And lastly, serendipity, because if you look for something and find something else that creates more value than what you were originally searching for, this is proof that we do not control everything and nor should we try to do so. Far beyond reaffirming the concept of free will, our society must cast aside its ability to map and control the future and start learning about trust and cooperation.
The fight is on for a SMART HR. There is an urgent need today for a consensus on good practice in the use of data and algorithms and particularly their application in the field of Human Resources. HR Big Data can create an enormous amount of economic value, but this will only make sense when it creates even greater social value in the long term. This is the topic of the recent theory by Michael Porter, one of the big names in capitalism, who explains that the aim of a company is not to maximize performance for its shareholders, but rather to optimize this performance over the long term for all stakeholders, whether they are shareholders, employees, or members of society. And what if HR Big Data could set this vision for the future to music?