Sales forecasting is not just a set of techniques that can accurately identify the volume of business for a given future period, or a method that can define business objectives at the beginning of the year; it is above all a success factor.

In Divisible, we have developed a "causal forecast" method based on correlation, starting from the assumption that the target variable, i.e. demand, can be described by the trend of other variables.
‍
In fact, product sales are influenced by numerous variables, both of an "endogenous" nature, therefore under the control of company management (product price, investment in advertising, etc.), and of an "exogenous" nature which may escape company control (per capita income, age of the population, interest rates, product taxation, rainfall, temperature, oil price, competitors’ moves, and so on).
‍
‍

❝ We use "simple regression" techniques to analyze the "causal relationships" between variables, if the demand is linked to only one variable,  or "multiple regression" techniques if the model considers two or more variables. ❞

In order to optimize the outputs of a sales forecast model it is essential, first of all, starting from precise and updated data and secondly, being able to predict the future values of the "exogenous" variables: this is the real competitive advantage.