A big part of a manager’s role is to make decisions. Making decisions comes naturally to most people although I have come across some who would prefer not to be put in a position where they have to decide. But just making a decision is not good enough – making good decisions is what separates good managers from bad ones.
The process of decision making has been much studied, with various hypotheses promulgated as to how it happens in real life situations. The truth is we make most of our decisions on the fly, all the time, whether it be crossing the road or deciding what television program to watch. Gut decisions might be best when deciding whether or not to run away from an escaped lion, but are extremely poor for most of the circumstances one is likely to encounter in the role of a manager. The trick is to know when to overcome the natural tendency to make a gut decision and instead use the information available, or determine what additional information is required, to make the optimum decision.
Sometimes we need to be able to make quick decisions but not “gut” decisions. In the mid-1980s, Gary Klein proposed what he called a “Recognition Primed Decision” model. He had interviewed Fire Commanders to try to understand how they made decisions. The Commanders were adamant that they did not make deliberate choices or consider alternatives or assess any probabilities.
In their situation a comparative analysis of all of the potential decisions that could be made, and a studied assessment of the most likely outcomes, is just not possible. If they followed that process, by the time the analysis was completed the building might have burnt to the ground and lives could have been lost. Rather, what they described was using their experience in similar situations to generate a plan of action, which they would then modify according to how things progressed. Once they knew what “type of case” it was, then based on prior experience they knew how to act. Without the luxury of time, this was the optimum approach.
Knowing what tool to use in each circumstance to aid decision making is a crucial part of management. Early on in my professional career I made a proposal for a certain course of action regarding a potential acquisition. In support, I backed up my proposal with facts and analysis and presented it to my boss. He listened patiently, nodding his head and then said, “could you produce a decision tree with all of the likely outcomes”. I went away and dutifully performed the exercise, thinking it would not make any difference to my proposal. I was humbled to discover that this simple tool showed that my proposal was flawed. My boss knew which tool to use in what circumstance.
Recruitment decisions are one of the decisions we tend to get wrong more often than we get right. Nearly all recruitment includes an interview process, even though the interview process is a notoriously unreliable predictor of who is likely to succeed in the role, or who the best candidate is. The interview process is like a poker game. The candidate is trying to show that they are the best fit for the role and that their experience makes them highly likely to succeed. On the other hand, the recruiter is trying to dig beneath the façade and discover what the person is really like.
Whether we like it or not, we are human, and we are not as rational as we like to think we are. We all have in-built biases. Studies have shown that even the best recruiters are influenced by things as simple as how the candidate shakes their hand or the time of day. And sadly, in this day and age, what sex you are still matters.
Management Guru, Peter Drucker thought that we get hiring decisions wrong because senior executives do not have sufficient discretionary time available to devote to the process and so it is shoehorned into an already busy schedule. That’s hardly ideal, when the senior executive should really get to know the candidates and understand their relative strengths and weaknesses. Time is the senior executives most valuable commodity; it’s a bad choice to spend precious little of it on recruitment.
I have a good friend who spent much of his career as an executive recruiter. His advice is to always look at how well the candidate had done in a similar role – leopards do not change their spots. Although reference checking is fraught with legal difficulties, he puts much score in frank discussions of a candidate’s abilities and performance in previous roles.
Too many years ago, I studied biochemistry. This scientific training taught me that to solve a problem I needed to determine what I wanted to know, come up with a working hypothesis, design and carry out an experiment to test this hypothesis, and analyse the results to determine if the hypothesis was valid or not. This works for some decisions but not all. For example, it works very well for testing the effectiveness of a marketing campaign but cannot easily be applied to slot machine purchases or recruiting.
Analysing experimental data is subject to bias, too. Studies consistently show that industry-funded research, whether it be an evaluation of the efficacy of a new drug or the environmental impact of a pesticide, are likely to present results more favourable for the client than that from truly independent research. It is not necessarily because the scientists are intentionally favouring their clients; it’s all too easy to let an unconscious bias creep in when decisions have to be made about the data.
Some problems are too complex for the human brain to get to grips with. Slot machine purchasing is a case in point. It is relatively simple to look at the slot floor and determine which machines are not performing well and are potential candidates for moving or replacing. But if they are moved or replaced what is the impact? What machines are the best to replace them with? Has the change increased revenue? Or has it only moved the revenue around the floor? Could the machine have been moved to a different location and made the floor even more profitable? These problems are non-linear, since moving one machine impacts the performance of all of those around it and impacts those where it was moved to, and probably the overall behaviour of the customers as well.
Fortunately, we now have plenty of computing power and smart people who can use artificial intelligence and machine learning to model the floor and predict the likely outcomes of machine moves or replacements. Using machine learning to “evolve” the mathematical modelling underlying the prediction sounds good, but how good is it? Operators need to trust the algorithm without really knowing how it works. And an evolutionary process does not always lead to the best outcome, humans being a case in point! Evolution can lead down a blind alley, as extinctions show.
Whatever the decision that needs to be made, a good manager will understand their own biases and know what information is needed and the tools to use to arrive at the optimal solution.
