We all want to make good decisions – however, doing so is often a challenge. The question in people analytics is how we can make decisions that are objectively good.
According to economic theory, the most efficient way of making decisions under uncertainty is to take a rational approach. With this approach you look at the probability of an event occurring, and the cost / value that you’ll realize if the event does occur.
You could define this cost / value in terms of monetary gain / loss, or as utility maximization. Utility is, in this case, the satisfaction experienced by the consumer for the good.
A rational actor will try and maximize on this value (based on what is often called a ‘loss function’). In simple categorical situations it’s possible to develop a decision tree to define clearly what is the optimal solution. In more complex situations you might identify the optimal solution via simulation.
People Analytics to date has mostly been focused on the probability side of this equation. I have argued in the past that People Analytics teams should spend more time building good loss functions on optimizing on these.
This is an approach and set of techniques that I teach to clients and are the basis of management recommendations. For most decisions that managers make in firms from the firm’s perspective this is probably the right way to do it.
Firms can absorb the losses associated with an ‘unlucky’ decision on the basis that the gains from ‘lucky’ decisions will outweigh them over multiple decisions.
Whilst this is true for firms it’s not the same for managers. One of the key findings of Kahneman & Tversky’s Prospect Theory is that losses loom larger than gains.
Empirical evidence shows that individuals tend to focus more on the changes in utility rather than maximizing overall utility. For example, if they can take a risk with a 50 / 50 outcome of either tripling their income or losing their job, they will most likely go for the safe option and not take the risk.
The way that incentives work in our organizations strengthens this behavior. For most managers, in most organizations, the long-term cost of making a bad decision (losing their job or just a derailed career path) far exceeds the potential upside of making a good decision (a small one-off bonus).
Hence the rational manager will take an approach of minimizing regret, not maximizing gain.
In many people decisions the expected value (e.g. the lifetime employee value) is realized over a long time basis. However the manager who makes the decision only benefits from this reward over a short time period (as they and their people move within the organization).
The negative outcome will have losses realized in a shorter time frame, again incentivizing short-term regret minimization rather than longer term utility maximization
This conflict between what is good for the organization capable of making multiple decisions and maximizing the mean expected gain and the manager who only gets multiple chances if they minimize regret leads to sub-optimal decision making.
A people analytics team modelling expected gain will always risk their recommendations staying on the Powerpoint slides. Being aware of the inherent conflict can help design strategies to minimize it.