ℹ️ The term "forecasting" is used in the case of chronological data and the term "prediction" in the case of data relating to a specified time period.
* * *
Predicting your sales will allow you to better meet demand, manage your stock or production capacity more accurately. Good forecasts can also serve your business policy: what is the impact if we increase our prices? What would be the impact of signing a customer today?
Predicting has become of crucial importance in our decision-making processes. Ideally, every business leader should have a forecaster to help them anticipate the course of events and take corrective action before problems arise.
The first question then arises is "Which method to choose? If the question is simple, the answer is generally very disappointing, even paralyzing. There is no better forecasting method than the others. Worse, they can all be correct in different cases.
Kopilot predicts the value that your sales could achieve. It is not concerned with the eventuality of the occurrence of a certain event, nor with the expiry of a certain event, i.e. the date on which an event that we know will have to occur will occur.
How kopilot made its prediction
kopilot offers a panel of prediction methods to provide a single prediction and a confidence index that allows you to better understand the different scenarios.
The prediction daily changes, and even intra-day, as business events occur.
Method #1 - Let's imagine you do exactly as Last Year for the remaining time till the end of the year.
Sales are recorded from the beginning of the year to today. From tomorrow, last year's values are added until the end of the year.
Method #2- Let's take the year-to-date growth rate into account to compute the sales for the non elapsed time.
Sales are recorded from the beginning of the year to today. From tomorrow, last year's values are added until the end of the year, but we apply the year-to-date growth rate to them.
This method is based on the assumption that the current growth rate will be maintained until the end of the year.
Method #3- Let's project the trend till the end of year.
This method is based on the monthly income of the previous 24 months. In other words, it will only be displayed if kopilot has sufficient time depth.
The current month revenue is replaced by the average income of the previous 24 months.
Next month is replaced by the average sales of the previous 23 months, plus the simulated sales for the current month.
And so on until the end of the year.
Method #4- Let's use some statistical modelling
The regression model seeks to establish a linear relationship between a variable, called explained (here the time), and one other variable called explanatory (sales).
This method is based on the last 36 months of sales. In other words, it will only be displayed if kopilot has sufficient time depth.