Demand Forecasting for a Colorado Microbrewery Using Decision Trees

Introduction and Business Context

A microbrewery in Colorado is growing rapidly. There is a long lead time for the purchase of new equipment, so management must make a demand forecast for the next couple of years in order to ensure that it has the capacity it needs to continue expansion. There are constraints, however, particularly with respect to access to key inputs. This casts uncertainty on the demand forecasts, such that simple extrapolation of current growth rates is insufficient. A decision tree is used to help with demand forecasting, taking into account different scenarios with respect to growth patterns and growth constraints.

The situation is that of a small microbrewery operating in Colorado. The microbrewery is five years old and has expanded annually since its inception. The brewery produces five beers regularly, available bottled for the packaged trade and in kegs for the hospitality trade. Several products are produced seasonally and are only packaged in kegs. The company's sales have grown to 10,000 barrels annually. The most popular product, the India Pale Ale (IPA), accounts for 50% of total sales by volume. The IPA accounts for roughly 40% of profits, because it has higher ingredient costs than some of the other products. The brewery uses a single brand for its different products, which are extensions of that brand. The IPA, because of its popularity, has been subject to further extensions in the seasonal range.

The signature ingredient of an India Pale Ale is hops, which are the flower of a vine. They are relatively difficult to grow, requiring specific conditions. They thrive only at certain latitudes in temperate regions and are cultivated in all great brewing nations, their presence being a critical requirement for the development of beer-drinking culture in the Middle Ages. New World nations like the United States, New Zealand, and Australia have cultivated the hop as well. The hop is relatively delicate and can be susceptible to disease, particularly if growing conditions are wetter than expected. There are different types of hop, and each strain produces flowers with different characteristics that impact the character of the end product.

Most hops used by the microbrewery are domestic varieties produced in Washington State, with some from Oregon and Idaho as well. Hops available to microbreweries tend to come from a handful of major wholesalers who source globally. They process the hops — usually into a pelletized form — for longer storage and easier distribution. Only the largest companies, like Anheuser-Busch, are vertically integrated to the point where they produce their own. Thus, almost the entire industry relies on these wholesalers, who dominate the North American trade.

The Operational Problem

The operational problem occurred when a confluence of factors resulted in a dramatic hop shortage across North America (Morgan, 2013; Welch, 2007). Adverse weather conditions produced poor harvests in two sequential years, reducing supply. Moreover, the growing popularity of microbreweries, and the hop-intensive IPA style in particular, resulted in rapidly increasing demand. The conditions of the wholesale industry favored larger breweries that were locked into long-term contracts with wholesalers. These breweries were allocated their hop needs. Smaller breweries, like the one examined here, which tended to purchase hops on an ad hoc basis, found that their orders could not be fulfilled. Orders would either be unfilled entirely or filled only on a partial basis.

For the brewery, this represented a significant challenge. It had limited supplies of hops on hand — for some varieties as little as two months' supply. It had growing demand for a flagship product that consumed 65% of total hop usage. With hop prices escalating due to supply issues, and in some cases supply being unavailable at any price, the brewery faced a serious forecasting challenge. The supply chain disruption was going to require changes at the marketing end. Even if the brewery could maintain IPA production, it might be forced to lower production of other products due to a lack of raw materials. Marketing would have to reduce dependence on IPA for growth, and that meant new product introductions. However, marketing was also facing a tide of consumer demand that would not accept mildly hopped beer, which for drinkers of microbrews is often equated with blandness.

Worse, the brewery needed to order equipment to facilitate its expanding business. If access to ingredients were not a problem, forecasting future demand would be relatively straightforward. However, since the brewery needs to order equipment now for delivery anywhere from six months to a year in the future, it needs a high-quality forecast of sales — including the possibility that it will not be able to produce enough IPA to meet expected demand for that product. Demand forecasting across a number of different scenarios is therefore necessary to give management the insight it needs to make the capacity expansion decision.

Forecasting demand had to take into account the trends of past sales growth, the industry trend toward IPA sales growth, and the effects of the marketing department's renewed emphasis on other products. These products would need to be scheduled based on expected sales and expected availability of the raw materials needed to produce them. Further complicating the issue is that the hop shortage also constrained imports from other parts of the world due to a global supply shortfall. In addition, market response to new product introductions is unknown. Forecasting is important not only because of the two-to-four-week lead time to produce beer, but also because of the six-month to one-year lead time for new equipment needed to maintain the brewery's pace of expansion.

Operations Management Principles

There are a number of operations management principles applicable to this case. The most critical is demand forecasting. The brewery needs a means of estimating future demand in order to set production schedules, which will in turn determine inventory management. The constraints on the availability of raw materials represent a particular challenge for this forecasting process.

Demand forecasting relies on both qualitative and quantitative information. On the quantitative side, a number of techniques can be used, including simulation and extrapolation. For example, the company is currently on a growth trajectory that can be extrapolated into the future. If growth is currently at 25% per year across all brands and the brand mix is not expected to change, then next year's demand forecast is simply this year's production multiplied by 1.25. For this brewery, however, that straightforward approach is insufficient.

Most forecasting also incorporates qualitative information. The base forecast is built on current demand levels, to which projected changes in demand are then added. For example, if IPA is expected to continue growing in popularity, it will grow faster than the overall portfolio. The qualitative analysis that identifies this trend is translated into a growth figure that is incorporated into the forecasting model. Similarly, if the hop shortage is going to constrain IPA production, that can be built into the model so that growth of that product is capped regardless of where demand might otherwise be headed.

A decision tree is a technique used to help a company analyze different possibilities. A basic decision tree might factor in low-growth, normal-growth, and high-growth scenarios, yielding three different demand forecasts. These can be weighted by probability to further inform the decision-making process. The brewery might further develop the tree by adding in probabilities for hop availability or demand probabilities for individual brands. Because the brewery is introducing new brands whose success is highly uncertain, the range of possible outcomes is wide. The decision tree therefore allows management to envision different possible futures and make better decisions as a result.

Conclusion

Operational problems are frequent in business. When a company is growing rapidly, there is often a tension between the pace of growth and the constraints on that growth. In many cases, financing is the binding constraint; here, it is the availability of a key input. Building demand forecasts is challenging under any circumstances, but when there is a long lead time for capacity expansion, such forecasts allow management to work within its financing capability to make the best possible decisions.

Good capacity decisions lead to more efficient production and higher return on investment, because the company has the capacity it needs to continue expanding without suffering from excess capacity. Identifying the key variables that will affect the demand forecast is critical; by acquiring and incorporating that information, the company can build a better forecast and therefore make a better decision. The decision tree technique is an excellent method for constructing such forecasts under conditions of uncertainty.

Morgan, J. (2013). Alaskan Brewing discontinues pale ale due to hop shortage. Craft Brewing Business. Retrieved July 24, 2013, from

Welch, D. (2007). Hops shortage likely to boost price of beer. NPR. Retrieved July 24, 2013, from http://www.npr.org/templates/story/story.php?storyId=16245024

Brewers Association. (2013). Craft brewing facts. Brewers Association. Retrieved July 24, 2013, from http://www.brewersassociation.org/pages/business-tools/craft-brewing-statistics/facts