demand forecasting

Much of our world has changed since COVID-19 took the global center stage in 2020. One aspect of the business world that has been challenged significantly is demand forecasting. The fragility of supply chains has been exposed not only in empty shelves in retail locations, but also in empty warehouses throughout the entire supply chain. As supply chains have been extended geographically, response times due to shifts in supply and demand have been strained at best.  In many cases, they have been shattered.

It would be reassuring if there were one set methodology we could employ that would make our organizations and supply chains impervious to “black swan” events such as COVID. However, no such process exists. At the same time, there are methods we can employ that make our organizations less susceptible to significant demand shifts. These methods are a mixture of quantitative and qualitative approaches that when combined, yield better, more robust results.

Before delving into specific approaches, we should agree on why demand forecasting is important to organizations. 

What is demand forecasting? 

Demand forecasting is the process of employing predictive analytics to estimate and project future demand for a product or service. At its core, demand forecasting helps organizations make better supply decisions by anticipating customer order volumes across time, markets and geographies. This enables an organization to make informed decisions about inventory management, logistics, and meeting customer expectations. 

Predictive Analytics Modeling for Demand Forecasting

Predictive analytics is often used in demand forecasting because these techniques can reveal patterns in customer purchasing. By seeking to forecast customer behavior using inputs such as historical data, business/market predictions and economic trend data, these models leverage statistical principles to provide a business team with data. The data is then used to create a framework for determining what factors influence consumer behavior.

Learn how Milliken can assist your organization with its unique business-tailored demand forecasting solutions.

Why is demand forecasting important?

At its core, the primary purpose of demand forecasting is to manage risk for an enterprise. Two primary risks are addressed: 

1. the risk of missing out on desired market opportunities
2. the risk of lost capacity or low asset utilization

Another important factor to consider is that forecasting is not an exact science. We will be wrong. Forecasting allows us to gain alignment and consensus regarding how the organization will respond as reality unfolds. With the overarching goal of managing risk in the face of some level of error, we can now consider several useful approaches.

Quantitative Methods

We begin here because every organization uses at least some form of quantitative demand forecasting. A business unit leader is asked to forecast demand for the coming year. One of their first questions will be, “What has the demand been historically?” Using historical data as the foundation for forecasting is a logical, long-standing practice. However, many organizations fail to mine all the useful information from demand history.  Some opportunities to better utilize historical demand include:

• Understanding demand patterns and seasonality: building models to better understand the ramp up and ramp down of seasonal products since this can have a significant impact on operational plans and inventory models throughout the year. Similar approaches can be leveraged when forecasting adoption rates and overall demand for new products that will cannibalize existing demand.


• Emerging/Diverging markets:  developing models looking at market adoption rates – those that are new/growing versus shrinking. These markets can be defined by geographies, affluence, ethnicities, and others. Failure to understand these changes can result in capacity and resources being misplaced compared to demand opportunities.

Although we want to understand long-term trajectories of product demand, the most recent data should take on the most significance in our forecasting models. The speed of competition and the falling barriers to entry in the global marketplace requires that we remain vigilant and continually review the current reality.

Qualitative Methods

If you're wondering what approaches might be helpful in forecasting demand for a new product and new market where parallels don’t exist or are not known to us, is a qualitative approach.  The characteristics of a product or product family that present such challenges include a fluid competitive landscape (low barriers to entry/exit, high rates of innovation), market extensions (existing products to new end-users, routes to market), and product offer extensions (product features, packaging, unitizing). Managing the risks discussed earlier is more difficult. One approach that many organizations use to aid in the effort is Scenario Analysis. This approach is done outside the core forecasting process. It involves the use of a team of internal and external subject matter experts who possess some level of knowledge about the product, market, and end-users. They develop likely scenarios for the market and score those scenarios on their likelihood of occurrence. As a result, the team provides proposed adjustments to the forecasts based on that analysis. However, more often the team will recommend a strategy for monitoring the market and layout possible responses, should specific scenarios begin to unfold.

MANAGING FORECASTS

To effectively manage forecasts over time, the organization must establish standards around two core activities – Measurement and Review.

MEASUREMENT

Measurement includes the inherent error in the forecasting process. Our concern is the margin of error and knowing we can accurately measure. One of the basic methods is Mean Absolute Percent Error, or MAPE. MAPE is popular for two reasons: 

1. it is simple to calculate
2. It is easy to understand

To calculate MAPE, the absolute difference between the forecast and actual demand for an SKU for a period is computed and then divided by the actual demand. The use of absolute value prevents over and under-forecasting from off-setting each other since an error in either direction can be detrimental to the business. Each business should assess its forecast accuracy and set a goal for MAPE. A solid starting goal for MAPE would be 10% by SKU. Where there are gaps in forecasting accuracy, this can be used to identify gaps in the existing process and improve forecasting in the future.

REVIEW

The organization must use a methodology for reviewing the forecast on an ongoing basis. To accomplish this goal, many organizations deploy a Sales & Operations Planning process (S&OP). The purpose of the S&OP process is to conduct short interval reviews (often monthly, sometimes weekly) of demand versus forecast so that the business can respond to significant differences. Those responses include, but are not limited to, modifying supplier orders, adjusting the schedules for internal operations, adjusting stock levels, and reallocating inventory locations. This is accomplished by engaging the key functional leaders who have the necessary resource assignment authority in a review to gain consensus on a near-term plan. This same structure can also serve as the catalyst to address more long-term forecasting issues on a less frequent basis (often quarterly).

Developing a robust demand forecasting process requires that we consider and leverage methods in four key areas. Our demand forecasting must include solid quantitative methods to extract as much insight from past demand as possible. Qualitative methods are also important tools; since history does not fully capture the reality of our current markets or any shifts in the competitive landscape. Leadership must measure the forecasting process to identify gaps and continually improve the forecasting process. Finally, the forecast and the plans that flow from the forecast must be reviewed regularly to allow for a timely response to unforeseen market changes. 

Milliken can assist organizations with demand forecasting, allowing them to make more informed decisions regarding their overall operations. We understand that client expectations are constantly shifting, necessitating a plan for efficiently forecasting demand. Are you ready to successfully forecast demand? Contact us today.