Forecast Methodology 3

[NathanLee]

Hi, Any suggestions on improving our manufacturing forecasts. Currently it is mainly based on word of mouth occurring between our sales guys (girls) and the customers. Can anybody refer me to good literature on the subject. Our forecasts seem to be plus/minus 50% off and I hope we can improve on this. Thanks for the comments!

__

Nate

 

[IRstuff]

The simplest thing to do is to compare against historical data. How do you know they're 50% off? It's because you have data, which means you have historical data.

A simple plot of data by month for year after year data should at least show if your business is cyclical and/or growing year after year.

TTFN

 

[PSE]

I agree with IRstuff to take a look at history (especially for patterns seasonal etc.) as well as to continue to communicate with the sales group. If/when they are successful in securing additional business, you can continue to plan accordingly. You can also begin to track how successful new accounts are in terms of actual vs anticipated placement of orders for new or additional product.

Regards,

 

[griffengm]

NathanLee,
One of the best places I've found is trade associations. They tend to track sales by industry against various factors. Once you understand what these are you can begin to compare your company's performance against some norms to see if the variation is internal or global.
Also check with them as to whether they look at "Best practices". Might get you started.

Griffy
It ain't necessarily what you know but how quickly you can find out who does!

 

[NathanLee]

So i've taken your advice, I already knew we were a cyclical business, however, we routinely overestimate right before the busy season. Next season we shall (hopefully) bouy our enthusiasm. Any recommendations on forecasting techniques for a cyclical business? Ie, literature I could go read? Thanks for your time!

 

[IRstuff]

The simplest is to use your existing historical data.

Since your business is cyclical, you can simply project historically by month, e.g., collect all historical data by calendar month and see where they're going.

TTFN