If you want to use Time Series; Past predicts the future objective approach. We would like to suggest you to use Exponential Smoothing Models. Considering your product seasonality.

PS:

1. Forecasts are always wrong

2. Aggregated forecasts are more accurate

3. Shorter time horizon forecasts are more accurate

Historic data may show hi/lo ranges from which you can generalise, and may also show seasonality. But there will also be many qualitative factors such as price sensitivity, what your competitors are offering, demand by your customer's own customers etc. Hopefully your sales force are on top of all these qualitative issues.

Definitely according to previous data.

Hello,

There are, to my mind, two main scenarios:

- either the customer is already known: in this case, it is necessary to study statistically the evolution and the regularity of the orders placed;

- either the customer is new: in this case, sales probabilities should be investigated based on a recent and reliable market study.

Bye

Demand

- Functional

- innovative

Supply

- Stable activity

- Dynamic Activity

Based on the above classification you can analyse the past data and come up with predictive pattern and train that data for analytical decision making.

Hello again ! There are, in fact, two main scenarios: - either the customer is already known: in this case, it is necessary to study statistically the evolution (mathematical regression) and the regularity of the orders placed - this allows to forecast, by correlation, future sales (by integrating seasonal factors); - either the customer is new: in this case, sales probabilities should be investigated based on a recent and reliable market study.

Is it clear enough? Bye