Whether it's Gap officials determining how many sweaters to stock on their shelves or a local diner making waitstaff decisions, businesses large and small rely on forecasting methods every day.
And experts say these forecasts can play a major role in driving company success, or failure, since accurate forecasting can help keep prices low by minimizing business losses.
"One of the ways you can be better than your competitor is not just by offering a better product," says Notre Dame Professor Barry Keating, an expert in business forecasting. "It's by forecasting better than your competitor does."
So you know how important forecasting is. But what are the best ways to go about it?
"When firms ask, whats the best forecasting technique? The answer is, it depends. Businesses don’t like to hear that, but it does depend," Keating says. "It depends on how much data you have, on what you’ve done in the past, and it depends on the kind of thing you’re forecasting and its importance."
There are two major categories of forecasting methods: qualitative, which typically uses expert opinions, and quantitative, which relies on historical or "time-series" data.
While its not a perfect science, there are some factors that can help you determine which forecasting method is best for your business. Here are eight questions to ask yourself to help you make that decision.
1. Do you have historical data available?
While most major businesses use quantitative "time-series" methods for forecasting, according to Keating, these methods can be used only if you have the data to draw from. So if you don't have historical data to use, you're likely going to want to turn to a qualitative method.
2. How much time and money are you willing to spend on your forecast?
This is another important question to ask yourself before you start diving into various forecasting methods, according to University of Pennsylvania Professor J. Scott Armstrong, since forecasting methods can range from those run on simple Excel plug-ins to the pricey assembling of industry experts.
3. Is your product or service a new offering?
One of the most common times that you won't have historical data is if you're trying to open a new business or launch a new product. If you're in these situations, there are a few strategies you can take, according to the experts.
If you're starting a new business, Armstrong recommends that you consider using the Delphi technique, a qualitative method which makes projections based on a structured survey of experts.
"You can draw a variety of expertise and have them all answer specific questions," says Armstrong, who teaches business forecasting at University of Pennsylvania's Wharton School. "That’s a useful method."
On the other hand, if you're an established business that's offering a new product or service, you might still be able to use a time-series method, Keating says, by using data from a similar product to make your projections.
4. What kind of product or service do you sell?
This question is especially important when trying to narrow down which kind of time-series method you should use. If you're a retailer that needs to take into account any sort of seasonal differences or other trends in your forecast, you should look into using the exponential smoothing model, a relatively simple technique.
"It breaks down the data into different patterns, and then it allows you to forecast that into the future," Keating says. "Does it work? Yeah, it works really well. It is definitely one of the most-used forecasting techniques."
The exponential smoothing model is especially useful for firms that need to make forecasts for lots of different stock units, he adds.
Businesses that offer only one or two main products, such as a utility company, should likely employ a more complex and time-consuming technique, such as one that uses a statistical regression.
5. Is your company being affected by the current economic downturn?
Unfortunately, a lot of traditional forecasting methods go out the window when the economy tanks.
"The federal reserve bank did a study last year of businesses forecasting. It turns out when an economy goes into the recession, like we just did, that forecasting errors of firms quadruple," Keating says. "The error increases 400 percent from what it normally is."
So what's a business to do? If your company has seen its business affected by the recession, Keating suggests the use of time series decomposition, a method that adjusts for trend and seasonality, but also for business cycles.
"If you use that method you can reduce your errors back down to a normal range," he says.
6. How many years of data do you have available?
The exponential smoothing method is relatively simple and accurate, but it also requires roughly three years of data. If you have less historical information, such as when attempting to forecast sales for a recently launched product, you can turn to a logistics model, which needs fewer data points, Keating says.
7. How far out are you attempting to forecast?
While time-series methods are often the most accurate, they are most effective for forecasting relatively near into the future. If you're attempting to forecast years or even decades out, you're probably better off relying on experts as opposed to statistics.
This is best illustrated by thinking of a hypothetical current attempt to forecast the percentage of students who will use electronic textbooks in ten years. "If I extrapolated out from the current data, I would tell you that it would be 6 or 7 percent," Keating says. "Now I'm pretty sure that’s wrong."
In such a situation, Keating recommends using a survey of expert opinions.
8. How important is your forecast?
All forecasts are important, but some are likely more make-or-break than others.
If you are making a major investment or taking a big risk based on a forecast, you will want it to be as accurate as possible. In those cases, Armstrong says that using multiple methods and taking the average result can help improve accuracy.
"You can hardly ever make things worse," Armstrong says. "You can only improve things. You can always reduce your risk."
Photo credit: iStock