HR analytics promise to give HR better data so they can make better decisions. And who would disagree? If you don’t know your levels of, say, talent turnover, how do you know if you have a problem, which of your current programs work and which don’t etc. In other words, with no or poor data you cannot make good decisions. I totally agree.
The incorrect assumption, which is sometimes made by some in the HR analytics community is, however, that people automatically will (or is even likely to) base their decisions upon good objective data. I would argue this is wrong.
I have previously argued that the concept of cognitive dissonance will actually make people make irrational and poor decisions even faced with good and objective facts. Elsewhere it has been argued that information overload (too much good data) will make people to make emotional or irrational decisions. Studies with placebo effects have also proved that people will act and make decisions based upon (irrational) beliefs rather than facts and data.
To understand why, I think it is important to remind ourselves of the difference between data, information and knowledge (see Davenport and Prusak for more).
1. Data is the fact of the world which can be subject to graphs, tables, statistics etc. They often reside in company’s archives, computers or records. Data do not carry any inherent meaning and more data is not always better as it can be difficult to make sense of raw data.
Examples: We have 5,000 employees, our annual talent turnover is 8.2%, profit per FTE (Full Time Equivalents) is $100,000.
Implication: Human beings cannot base their actions or change behavior on data alone!
2. Information is inferred from data and essentially means that the data has been processed by a human being (or increasing by an intelligent computer) and conclusions are drawn on the basis of the data. Peter Drucker said that information is “data endowed with meaning and relevance”. Some call them “value-added data” – it is sent from sender to receiver indented to change the receiver’s perception of something.
Examples: The talent turnover rate is too high and should be at 4%, our talent turnover is higher than our competitors’ and have been trending upwards
Implication: Evidence from diverse areas such as behavioral finance, social psychology and neuroscience show that people rarely act on the basis of information. Concepts such as cognitive dissonance (and many other) simply override information. More habit-breaking actions are impossible to make on the basis of information whereas more routine and simple actions can be done.
3. Knowledge is inferred from information and is produced by taking information and adding experience, evidence (research, case-studies, theory), contextual information, consequence etc. To produce knowledge requires human beings – it cannot be produced by intelligent software only. It represents our ‘map of the world’.
Examples: Increasing pay levels will not reduce talent turnover in our company but a 2-year talent management program will.
Implication: Knowledge is the only level which will produce ‘real decisions’ and therefore impact behavior in individuals and organizations. HR must take the best possible information and turn it into knowledge before it has the impact it was meant to have. An approach which looks at a mix of experience, values, context and not least (a high degree of) evidence-based research will produce knowledge.
So when we say that HR analytics will enable HR to make better HR decisions we need to understand that it has the potential but it will not necessarily do so. If the data is presented as, well, data then it will have little or no effect. If it is produced as information it may have some but not much impact. If HR analytics can produce real knowledge it will have a profound impact.
The solution is therefore not to remove HR analytics away from HR as has been suggested by some. No doubt that the actual skills of setting up and running an effective analytics department requires skills not present in any HR department today. But removing it too far from HR makes it difficult to use this data and information to produce knowledge. Knowledge is produced by human beings with experience in the subject matter. At least a (very) strong connection between the analytics department and HR must be established.
I freely admit that I believe HR analytics can add a lot of value in any organization. I also strongly suggest that it is only by converting information to knowledge that it can fulfill its promise of being an enabler of better HR decisions.