I'm a data scientist and look into questions like this frequently. Until recently, I ran the data science organization at a payment processor, where we answered questions like this for merchants on a daily basis, and also built products at scale to help inform merchants' decision making.

The question here really is: can you predict these sales numbers well enough to inform specific buying and manufacturing decisions? Sales numbers are difficult to predict precisely for many reasons (week to week fluctuations can be pretty high, seasonality can be a big factor depending on what you're selling, it's very helpful but often difficult to separate sales from existing and new customers, etc.), but as long as you can predict them well enough and far enough in advance to make better decisions you can create a positive financial impact. The point is that this isn't so much a question about predicting the sales as well as you can, but a question about optimizing your decision making process.

Therefore, I'd encourage you to engage the help of someone who has experience in solving operational business problems using data. One of the first things a good data person should do is help you quantify the size of the problem - how much money you can save by doing this analysis. This will help you determine how much time and money to put into it to make sure that the project has a positive return on investment.