Forecasting: Relying On The Unknown

By Amy Radishofski, Staff Reporter, Manufacturing.netDespite being busiest only during certain months, seasonal manufacturers need to forecast and make decisions that can make or break their entire year. Proper planning and collaboration are paramount.


September 23, 2007 was the autumnal equinox — the official end of summer. The leaves are beginning to turn colors and the seasons are changing.
 
For some manufacturers, the end of summer could be the end of a busy season, or it could just be the beginning. If you manufacture beach towels or Christmas trees, you’re locked into seasonal demand.
 
How do you adjust your operations to handle your peak time? How much of a peak period will you have? How much capacity will you need? What will be your inventory storage and holding costs?
 
Factors and Consequences
 
There is a lot of variability when it comes to demand — consumer tastes may change, competition could increase, weather patterns could change, etc.
 
“The farther away in time a forecast is from the sales it projects, the less accurate the forecast will be. This stands to reason that the longer the horizon, the more changes will take place between the forecast and the actual sales,” said Jane Lee, Vice President of Supply Chain, Supply Chain Consultants.
 
“Picture a company that makes orange juice,” said Jim LeSage, Executive Vice President with The Facility Group. “They may be producing more at the start of the school year or during cold and flu season, but the oranges are only harvested at a certain time.”
 
That leaves the orange juice company with few options, LeSage explains. The company can store the finished product or it can store it in working form.
 
Even if you’ve checked all the factors and made the best prediction you possibly can, there is no guarantee that you’ll be right.
 
“Remember the Atkins diet? Think about the number of companies that were impacted by its unexpected popularity,” said LeSage. “While you can’t predict everything, you can hedge against it.”
 
When is the best time to do your forecasting?
 
“It’s a truism of forecasting that the higher (more general) the level of the forecast, the more accurate it will be,” Lee said. “It is much easier to project total automobile sales 18 months from now than sales of red Toyota Corollas.”
 
“Your best forecast is your actual order,” jokes LeSage. “But you should delay it as long as possible so you can get a clearer picture of what’s going on around the product.”
 
You can look to the past for forecasting, but situations can change dramatically. Your marketing department can be running special promotions, or there can be packaging changes that can delay final production.
 
Short, Medium and Long Term Forecasting
 
“I’ve had success breaking up plans into short, medium and long term periods. Each of these durations needs to be defined for the particular business you’re dealing with,” said LeSage.
 
“The key is in determining for each business what is acceptable as ‘long term’ versus short term. For making automobiles, a three week detailed forecast may be fine; for most process industries, however, which have long production cycles, complex and time-consuming transitions based on product sequence must be dealt with for any adequate projection of capacity,” said Lee. “In these situations, there is no way to project capacity without projecting mix and sequence to some degree. It is therefore necessary for most process industries to do detailed forecasting for a much longer horizon than the white paper suggests.”
 
A Supply Chain Consultants white paper, “Forecast Less And Get Better Results,” identifies a point in the future — the Planning Time Fence — which is key to short term, detailed forecasting.
 
“The Planning Time Fence is the horizon within which a business needs to forecast at the detailed level. It is determined by a combination of how far in advance raw materials must be planned and how far in advance detailed production capacity must be known,” Lee said.
 
Outside the Planning Time Fence, companies should deal with aggregate volumes. The white paper also warns against the drive to use technology “just because you can” and to stay away from the Suicide Quadrant — the use of detailed, long-term forecasting.
 
“Detailed short term forecasts are essential for planning and scheduling production,” Lee said.
 
“Medium term plans can be useful in understanding things that are a little more flexible to transition but don’t have ability to turn on and off like lights. Planning for staffing and crews, holiday promotions that have long lead inventory build requirements, graphics, changes, etc,” LeSage said. “It is especially useful to have good aggregate long term plans for capital expenditure planning, production planning with transportation/logistic considerations, product line declines and transformation of capabilities to new consumer needs.”
 
“Seasonal manufacturers face a different set of challenges. It is likely that they must produce much of their seasonal demand far in advance of the season because they lack the instantaneous capacity to produce what they sell each month during the season,” Lee said. “In this case, forecasting by month is less important than forecasting correctly the volume needed for the entire season.
 
“On the other hand, if inventory is to be pre-built, it must be done at the SKU level, so knowing the seasonal need at the SKU level is essential. One strategy for effectively approaching these challenges is to pre-build only a certain percentage of the expected season’s demand for any given SKU. This minimizes the risk of guessing wrong on the mix and having way too much of some SKUs left over at the end of the season.”
 
“Forecasting has a collaborative nature,” LeSage adds. “Solving issues should be done in a collaborative manner across the supply chain, sales and marketing organizations.”
More in Home