Managers from different companies, industries and countries recurrently express their disappointment with sales forecasting accuracy. The traditional approach to forecasting -which is composed of worst, most-likely and best case scenarios- isn’t lousy because it is wrong. Its inaccuracy is based on a wrong approach that aims to be too right, too precise.
Traditional sales forecasting creates the illusion of precision, which does managers a disservice as they need truthful forecasts to plan their companies’ operations. Remember that meteorologists, for instance, are scientists with a huge amount of information coming from different sources, sophisticated arithmetical models and yet they can’t forecast the weather with total precision.
There are dozens of variables involved in a sales decision process, and the sales force can’t manage all of them. So they focus on the most important ones they can affect and pretend as if they are 100% responsible for that closing.
The use of Montecarlo as a screening tool comes from the work that was done while developing the atomic bomb during World War II. In the initial stage of this research, John Von Neumann and Stanislaw Ulam defined this method, inspired by the results of a Montecarlo casino roulette gambling.
To understand the Montecarlo simulation, first consider the case where a simple and linear event needs to be predicted. For instance, when you need to calculate the direction and velocity of a baseball, you can use a linear equation to determine how far it will fly. This is a deterministic case in which identical initial conditions will always lead to the same result.
However, more complicated systems require complex interactions among many variables. The Montecarlo simulation uses random inputs to model the system and produces a probable outcome.
Montecarlo helps us to emulate realistic world combinations of independent variables and provides us with a reliable probabilistic result. This implies a huge shift in the way we think about sales forecast. Instead of thinking about a specific number, we are now thinking about the probability of a defined range of possible results occurring.
For instance, instead of considering that sales for the current quarter are going to be 55M, we should think that there is a 70% probability (close to one standard deviation) that the quarterly sales will be between 35M and 75M.
As we come to the end of the quarter, the degree of uncertainty decreases and therefore the bell curve narrows and consequently the expected results will be more accurate.
Instead of providing a single number, we get probabilities for diverse potential sales that could be used to more accurately plan the company’s operation.
The Montecarlo simulation is a stochastic technique, which means that it’s based on the use of random numbers and probabilities to estimate real life circumstances. It provides decision makers with a nuance of possible outcomes and the probability that they will happen.
The aim of this method is to determine how many random variables affect the performance of the system that is being modeled.
Probabilities and ranges are not as pleasant to use as a specific number, and they require a real shift in the management mindset. Few people feel comfortable accepting that the future is uncertain. The human being often creates different mechanisms to give a false sense of security to an unsafe environment. Accepting that the future is undefined and planning based on the probabilities of potential results, leaves us in the best position to maximize those results.
Part of this process of acceptance means that sometimes the results will go against the odds. But, consistent decision-making based on probabilities will finally provide the best results in the long run.
One of the greatest management roles is to drive the business through an uncertain future. Executives make decisions that will determine the direction of their companies, in most of the cases with partial information. Let’s ensure we are making our decisions based on the most accurate information possible.
