June 26, 2015 | By Greg Decroix | Back to blog

Supply chain managers are constantly seeking better information about future product demand to better manage purchasing and production decisions.

In some cases, companies are able to easily obtain specific information about demand—for example, in industries like commercial airliner manufacturing where production backlogs are the norm, customers place orders with the company in advance of when the items are actually needed.

In other settings, future demand information may contain more “noise” and may not provide specific information on individual products but rather may contain aggregate information about demand across multiple products.

Gregory DeCroix
Gregory DeCroix, Professor of Operations and Information Management at the Wisconsin School of Business

Fernando Bernstein of Duke University and I explored the use and value of this type of less specific, more aggregated demand information in settings where multiple products are being produced using multiple resources (e.g., components or production lines). We focused on two particular types of aggregated demand information that arise naturally in a variety of settings: the total demand volume across a family of related products (or across different geographical regions) and the mix of demand across products within such a family (or among geographic regions).

As an example of volume information, consider the auto manufacturers that were forced to change some of their paint colors due to pigment supply disruptions arising from the 2011 earthquake and tsunami in Japan. The mix of demand across these new colors was rather uncertain, but if demand for a model is reasonably stable year to year, the previous year’s sales may provide a good, though still imperfect, signal of total demand volume across the color options.

Companies can sometimes obtain volume information through the design of their production capabilities and/or the way they work with customers. For example, by allowing retailers to change the color mix of their initial product orders and by incorporating postponement concepts into their production capabilities, the Italian clothing retailer Benetton has been able to obtain advance information about total demand volume prior to the selling season while (color) mix information is still uncertain.

Firms may be able to get demand mix information from surveys that assess customer preferences among different product variants. Purchasing quantities reported by customers in such surveys can sometimes be somewhat inflated, limiting the amount of volume information that can be obtained, but the relative preferences revealed by the surveys (mix information) may be more reliable.

Consumers’ online behavior can be another source of mix information. A company we spoke with sells belts and other accessories and can obtain clickstream data on the number of times each product variant is viewed. Such data can provide some sense of the relative popularity of the various belt designs (mix information). Because conversion rates in online commerce tend to be low and volatile, however, such data may not be that useful in predicting demand volume.

Finally, a company choosing to enter a new product market (e.g., tablet computers) may be able to analyze incumbent companies’ sales to estimate customer preferences among different variants (e.g., different memory levels), but this information might not provide much guidance as to the overall strength of customer response to the new company’s products (volume information).

While improved demand forecasting is one way to deal with an uncertain future, another strategy is to invest in production flexibility—that is, to develop flexible production lines or shared components that allow the company to respond to possible future scenarios as opposed to predicting which will occur. In order to explore potential synergies (or conflicts) between these two types of investment, we modeled and analyzed a firm under different combinations of production resource capabilities (distinct resources dedicated to each product or one flexible resource) and information type (mix or volume).

We were able to identify some interesting interactions between resource and information type, with certain combinations pairing naturally. For example, volume information and resource flexibility are complementary—in other words, volume information is more valuable when a company has flexible production resources. (The opposite is also true—if a company has access to volume information, investments in resource flexibility deliver more value.) On the other hand, mix information is less valuable when resources are flexible than when they are dedicated. This makes intuitive sense—with production flexibility, it is less important to predict in advance which products will sell, but it is quite valuable to know how many total units will sell.

We also found that the resource type affected whether mix and volume information reinforced each other or could be viewed as substitutes. With dedicated production resources, either information type becomes more valuable in the presence of the other, but the reverse is true when the production resource is flexible.

Taken together, these various insights can help guide managers as they weigh potential investments in either of these types of strategies—information or flexibility—for dealing with future demand uncertainty.

For more on this topic, see our paper “Advance Demand Information in a Multiproduct System” in Manufacturing & Service Operations Management.

 

 


Categories: