Guessing, justifying, or deciding?
When making decisions in companies, those responsible find themselves with a great difficulty: the impossibility of knowing what will happen in the future. Some challenges that may arise depending on the approach adopted and how we improve our decisions…
Simulation for decision-making
The problem of uncertainty is a complication that man has been trying to solve since the ancient times of humanity, having come a long way from black magic, reading coffee grounds or consulting oracles or crystal balls.
When faced with uncertainty, executives often have extreme reactions. Some simply pretend that the uncertainty does not exist, while others see it and are paralyzed by its presence.
If we cannot know it, why not guess it?
For some time now, the forecasting technique has been used in companies to make long-term plans and decisions. Regardless of the path traveled, forecasting continues to attempt to be a technique that allows us to predict what will happen in the future.
Business decisions are always made with insufficient information and with a greater or lesser margin of uncertainty, depending on the time and resources allocated to the search and analysis of information. Faced with this reality, the analysis through forecasts supposes that, since determining with accuracy cannot be achieved, some assumptions must be established and acted upon accordingly.
The forecast is usually expressed as a single value, or as a set of data. In turn, this set could be expressed with two or three possible values, called “scenarios”. For example, I can propose a pessimistic, a middle-ground and an optimistic scenario. The sales planner needs to have an idea of the amount of inventory to be produced to meet future market demands. The forecast value is usually derived from an average demand value taken from a sample of historical data or is simply the best estimate of those making the forecast.
In most cases, we know things will not happen as predicted, so then, why forecast? One reason may be the lack of knowledge of other methods, and additionally, because a basis is needed (even if it is minimal), a justification for the course of action that is proposed to be taken. Because from the forecast that we generate, we will be deciding where to allocate our financial, technological, and human resources. Thus, forecasts are subjected to tendentious pressures, that often push them further away from reality.
What can go wrong with this approach?
The use of averages distorts business decisions from day to day. Let’s imagine a product manager for a new technology who is asked by his boss to forecast demand for his product. “That’s very difficult for a new product”, replies the product manager, “…but quite possibly, the annual demand will be between 50,000 and 150,000 units.” “I need you to give me a number to give to the production people.” “If you need a number, I think the average of 100,000 is going to be representative”, the manager says hesitantly.
If finally, the demand is below average, clearly the benefits will be lower than estimated. But, if the demand surpasses the average, it exceeds the capacity of the plant, the maximum profit will remain stuck at the estimate for the average.
Thus, the use of averages ensures that the average benefit will be less than the benefit associated with the average demand! This leads to the product manager being blamed for not achieving the expected benefit, even though the estimated average demand was correct.
In the areas where executives make decisions today, the impact of uncertainty is becoming more and more important.
For many variables, it is not possible to have representative historical data (eg, protectionist policy measures), and for those for which information is available, it cannot be used as a good basis for forecasting the future due to the huge differences between past and current circumstances.
The existence of these high levels of uncertainty is not intended to encourage a position of helplessness by arguing that because a result is uncertain, it is not subject to systematic and conscious analysis; but on the contrary, to account for new analysis methodologies such as the stochastic simulation of scenarios.
It is necessary to create a culture that rewards a statistical and probabilistic way of thinking over the intuitive one.
Simulation, in the most common sense of the word, means to imitate. And this is precisely what it is about: the behavior of a system is going to be imitated through the manipulation of a model that represents a reality. The advancement of software programs enables us today to develop models for business decisions.
Executives understand that the explicit treatment of uncertainty when making decisions improves the communication of complex issues, promotes the generation of contingency plans reducing surprises, allows to know the opportunities of potential benefits and explore the factors that can produce extreme results.
In order to enjoy these benefits, it is necessary to create a culture that rewards a statistical and probabilistic way of thinking over the intuitive one. And here we are not referring to the knowledge and ability to handle figures and sophisticated calculations or computer programs for statistical analysis. It is simply that people must think that reality is uncertain and as such, their results are variable.
Partner at Tandem.