Comments on: Where Should We Place the Forecasting Function?

By: Rick Kesler

Rick Kesler — Wed, 09 Sep 2009 20:32:57 +0000

Good discussions … however, as a consultant in this topic, I feel I have best answer. First we must define a few things.
Demand forecasting is resposible for colsolidation of forecast inputs from all business unit and creating an unconstrained forecasted using those inputs, historical data, and any number of demand programs.
This group can reside within the supply chain but MUST not be tied to any of the business unit (free of all bonus targets and political interferences) KPI is forecast accuracy (both consolidated and from the business units). They manage the concensus process and provide the business units with feedback on gaps. They also have finace budgeting responsibilities (NOT MANUFACTURING BUDGETING)
This unconstrained forecast can then be given to the supply planning department (also within the SC). That department manages the link between the inventory (SS, WFC, OOS, backorders, etc..) and the production site via MPS or MPP. They use site manufacturing constraints, product DC balancing, customer proritixation matrix, etc… to produce the constrained supply. The are the communication hub between the commerical group and the manufacturing group and convert the forecasted budget into a MANUFACTURING budget.
The S&OP process is managed through the operations group which usually reports through the finance department. This group manages sku lifecycle, new product launches, ROI, cost, etc… The S&OP process is NOT a concensus meeting! It was origially designed as a communication tool to discuss short term production issues and long term capacity constraints. Demand forecasting, supply planning, marketing, sales, finace, and operation all present current KPI’s and stratigic visions. The manufacturing sites also participate to share capital plans as well as to understand future product budgeting. Feel free to shoot me a question if I am unclear (rubygem73@yahoo.com)

By: Chaman L. Jain

Chaman L. Jain — Thu, 03 Sep 2009 23:48:21 +0000

Chris

I did not imply that companies revise their forecasts too frequently. I just gave the numbers what companies are doing. It is a waste of time and money if the revised forecasts do not add any value.I think forecasts should be revised when sufficinet amount of additional data becomes available, market dynamic has changed for one reason or the other (one competitior has left the market or new one has entered the market, competitor has recalled its product, which will cause an increase in the sale of your product, or for any other reason),or product has an extremely short life cycle. Even if we improve our forecasts with frequent revisions, but we don’t have enough time to react, it would not help. To decide how often to revise, one should determine whether the incremental cost is worth the incremental benefit, which may vary from product to product, company to company, and industry to industry.

By: Chris Schinzer

Chris Schinzer — Thu, 03 Sep 2009 23:22:39 +0000

Chaman,

Thank you for taking time to answer my question. I now pose a new question to you. It seems that revising your forecasts more often would provide more accurate predictions, true? So wouldn’t a company be better off to revise every 2 weeks, or every week, or even so much as every day? And lets say that this is true, and I provided a company a method in which to revise their forecasts more often, then a company would be able to very accurately predict their sales after time. Couldn’t this save everyone in the supply chain a great deal of money in warehousing costs?

Thank you for your time Chaman,

Chris

By: Dan Rasoi

Dan Rasoi — Thu, 03 Sep 2009 00:20:21 +0000

The forecasting department should reside within the the department that is responsible to manage inventory or production.

Three reasons:

1. The department will be measured on how well they service customers as well as Inventory levels.

2. Credibility and trust. If you would be charged with managing inventory, would any of you be more inclined to trust and accept a forecast from a employee from a different department that may have different goals and objectives?

3. Understanding of forecast error. The sign of the forecast-error is key to managing inventory, it should be the only measure for organizations managing physical inventory.

Unfortunately many organizations use MAPE which is worthless, MAPE does not give an indication of over or under-forecasting. MAPE is easy to report corporately and many corporate managers like it just because it is easy to compile and report. Another horrific fallacy of MAPE is that it measures how you performed against the Actuals, not what percent of the Forecast (plan) did you actually achieve. Let’s say your MAPE is 10% however the past three months your Sales/Finacial Plan underperformed by 6% and you have been building inventory?

By: admin

admin — Fri, 28 Aug 2009 20:56:23 +0000

Hi John, I agree with Robin’s approach. Forecast ownership has always been argued as either belonging to the supply chain or to sales/marketing. My push is that it it belongs to both. Demand and supply balancing can’t be done well in silo’s.
The approach I am trying to implement is basically; demand management creates a forecasted range of possibilities based on history, trends, regression models, etc. These forecasts are then given to sales/marketing with the qualifying assumption that these forecasts will be accurate as long as historical sales are a good indicator of future sales. Sales then identifies where this assumption is not valid (new competitors, new markets, recession, etc) and over-writes the baseline forecast while documenting the assumption changes. Sales also over-writes the baseline for any events (marketing campaigns, channel loading, etc). Measure the accuracy of both the analytical and the marketing over written forecast and publish forecast value added metrics. Use the FVA metrics to identify where sales needs to focus improvement efforts.
Supply chain could probably own building the analytical baseline forecast, but sales/marketing would need to own the FVA metrics (assumption and event management) to drive improvements in accuracy. S&OP would be a good place to manage this process.
Hope this helps?

Posted by Richard Herrin

By: Nicolai Dortmann

Nicolai Dortmann — Fri, 28 Aug 2009 18:38:37 +0000

Chaman,
in my opinion ownership of the forecast figures is key – departement which is closest to the area of what is forecasted to make forecast – but also be responsible and made liable for any deviation. Deviation in positiv as in negative to be seen as bad.
=> if you are not measured against your accepted target you will always have an excuse.
Who should better forecast sales then sales people / production better then manufacturing => but make everybody liable for deviation.

Regards

Nicolai

By: Chaman L. Jain

Chaman L. Jain — Fri, 28 Aug 2009 15:03:38 +0000

Chris

If I understand your question correctly, what you are asking is how often they revise their forecasts. Below are the results based on the recent survey by the Institute of Business Forecasting and Planning.

a. Once a month —64%
b. Once a quarter –13%
c. Once a year -3%
d. Others -20%

Specific events certainly have a role to play. If they are known at the time forecasts were prepared, their impacts are incorporated in forecasts. If they are discovered at the last minute, then their impacts are incorporated into the baseline forecasts intuitively and collectively in the monthly consensus meeting.

By: Chris Schinzer

Chris Schinzer — Fri, 28 Aug 2009 05:16:31 +0000

Chaman,

I am an economics major at the University of Michigan. I have just one question for you and that is, how often do suppliers update and/or forecast their demand function? Let’s say I had an idea that could update it up to the second and break it down in local markets, even as specific as a sporting event. Would that be of great value to producers and retailers?

Thank you for your time,

Chris.

By: admin

admin — Thu, 27 Aug 2009 13:54:56 +0000

Hi John, I’m the Demand and Revenue Planning Manager for innocent drinks and we’ve actually gone a 3rd way. We’ve setup a business delivery function that develops our market strategy and plans as well as predicting future demand and our Demand Planning function sits in that area. We get our account managers (sales guys/girls) to forecast their accounts and then have short term operational overrides in place from our operations teams based on input from customer’s operations teams.

Give me a shout if you want to talk more.

Cheers.
R

Posted by Robin Parsons

By: admin

admin — Thu, 27 Aug 2009 13:54:42 +0000

LinkedIn Groups
Group: Sales & Operations Planning Network
Subject: New comment (4) on “Where Should We Place the Forecasting Function?”
John,

One of the key ways to improve forecasts is to make sure at whatever level in the company is forecasting it is being measured.(i.e. by salesman, or by region, or SKU). Two Chemical companies I have worked with forecast SKU/major customer for A items and new product intros. By managing the A items you are getting the most cost crucial SKU’s that tie up a large percentage of your inventory with your time available. They still look at B’s and C’s just not as often.

Demand Managers are the most common assemblers of collaborative processes, and in some cases – because of limited resources, they just forecast SKU level by exception 80/20 themselves.

Normally the function is within sales and marketing, but I’ve seen it gravitate towards operations if no formal supply chain group exists.

Good luck in helping your company determine what people want and when they want.(true demand forecasts – customer req’d date history)

Posted by Mark Thomas