People pursue in a career in Human Resources for a number of reasons. Perhaps like myself they are drawn to it as they see the enormous potential impact it can have on driving the performance of a business. Perhaps they enjoy the interaction with people (a people person, probably the worst reason to get into HR and my pet peeve). Or there is a third common reason and that is this was an area that leveraged their strengths and didn’t expose their weaknesses.
Up until fairly recently to ensure a HR professional was a high performer, a high numerical aptitude was not a competency that was normally assessed during an interview or selection process. This part of the typical HR practitioner’s skill set didn’t really matter and for good reason, it simply wasn’t necessary for us to deliver against the common goals and challenges within HR.
Well, there is a shift happening. Specifically, a shift in expectations from the “C Suite” and it’s going to require HR professionals employ a level of numerical aptitude that’s not normal for them. Some will adapt, like many did to the use and introduction of technology and some of our HR colleagues won’t.
The shift I’m taking about is the application of predictive analytics. The opportunity to shift the way in which we make critical decisions from being largely intuitive to using intuition to form hypotheses that are validated by data driven approaches or predictive modeling.
At present there is a religious debate between using a intuitive approach when deciding on critical HR interventions, to an approach that also includes data driven techniques and modeling. Critics of predictive modeling simply believe that there are too many elements or variables to account for what may impact the evaluation of a particular strategy or approach.
In the fields of Finance, Insurance and Marketing, predictive analytics has taken massive steps, HR has taken baby steps. However in recent times we have started to realise that our profession can benefit greatly from this very same discipline. Within this decade predictive analytics will be a required activity in every HR environment.
Show me the evidence
There have now been hundreds of studies that prove that simple statistical prediction from an optimal regression equation is much more accurate than intuitive prediction alone. We, humans still need to come up with the hypotheses to be tested, but instead of intuition being the end of it, intuition is just the beginning of the equation, with predictive modeling testing potential interventions.
The most notable of studies included one by Meehl and Grove where they conducted 136 studies into the accuracy of judgment of experienced managers. Initially it was found that in all but 8 cases the power of predictive analytics significantly beat the manager’s intuition. So you’re probably thinking, well it least humans won 8 times, well it was found in a post study review that these 8 loses were due to random sampling errors. Human error. So a regression equation beat the manager every single time.
Another piece of notable research shows that high performing businesses are much more likely to view predictive analytics as a core capability than a low performing competitor. That those environments that embrace predictive modeling within their decision making are yielding big rewards. In the study “Strength in Numbers” by Brynjolfsson, Hitt and Kim in 2010 of 179 large companies, those adopting data-driven decision making were seeing productivity gains of 5 to 6 per cent higher than those of their competitors that were not. Although the percentage gains are commonly higher for small to medium organisations, 5 to 6 per cent is a massive chunk of change for a multi billion dollar fortune 500 company.
There is a long way to go for predictive analytics to be used in the same manner and sophistication that its used in other functions like marketing, sales and finance. However it’s clear that how we are going to make decisions is going to change over this next decade and it’s effectively equivalent to a change in mindset moving from “I think” to “I know”.
Despite what many may think, we’re not infallible
Science has proven that humans aren’t hardwired for making complex decisions that contain a variety of variables. We are not only prone to make biased decisions but we are typically overconfident about our predictions and slow to be convinced otherwise. This overconfidence becomes more acute the more complicated the prediction.
So take a bottle of Coke for example, we can simply predict that a not yet opened but shaken Coke bottle will explode in a mess when opened. Yet when the number of factors increase, we can’t cognitively assess how to weight and calculate these individual factors. These variables become too much of a cognitive load for a human brain to dynamically assess and interpret. For example, say the Coke bottle has been opened, is more than half finished and has been sitting in a hot car for the last 4 hours with the top only 80% closed. Its much harder for us, even experts to weight all these variables, calculate the amount remaining, by the temp of the car, by how hard the top has been shut to determine of there is any carbonation left to fizz up.
Predictive modeling in HR focuses mostly on finding predictive patterns of employee turnover, performance and workforce planning. Forward looking, it combines algorithms, historical information and data mining to solve problems, realize an outcome or answer a question. Such as the probability an individual or a sub group of employees is likely to resign from a position. Or what mixture of skills, experience and competencies would most likely guarantee a high performing new hire or promotion. With this information, analysis can be applied to predict how successful different courses of action will be.
So can a HR practitioner master predictive modeling?
Like learning anything new, you will stumble and fall. What is important is that you pick yourself up and try again. In fact much more important than intellect is the need for tenacity, just like a boxer the ability to take ten punches to the stomach and get up for the 11th and being prepared that the risk you take and the failure you inevitably face is exactly the right path. Failure is the route.
Those that persist, will find mastery and those that don’t, won’t. In fact failure is part of the predictive modeling process. In order for a predictive model to work, you must first test your intuition and in many cases, this intuition will not be correct or the model may not be quite right, the data you use may not be of the right quality or quantity, so you may need to tweak your approach. In other words the key is to fail fast, apply key learnings and iterate. The results will speak for themselves.
It all starts with collecting the right data. At smarterhire we validate the feedback received by our leading exit interview and probation review products with historical data to ensure the validity of any patterns or trends and use predictive modelling to help identify the best intervention. We provide people intelligence informing organisations about insights about their teams and individuals as well as actions or interventions our clients can take to produce a better outcome.
Predictive Analytics HR
