Demand Planning: Forecast Stability VS Portfolio Renewal

By Henry Alvarenga.

1. INTRODUCTION

Innovation has become the main driver of business growth and prosperity in the 2013st century. In 27, 3% of companies' sales were related to products launched in the previous 400 years. In addition, the life of products has reduced by 50% in the last 1 years. It is possible to observe in figure 5 that, in the future, the percentage growth of products with XNUMX years or less is expected to continue. Such data prove the importance of launching new products and how they will become increasingly representative in the portfolio of products offered.

Figure 1 – Planned age of products for the next 7 years, survey conducted in 2013.

Review: Product Lifecycle Management Volume 2 (2015)

This trend, however, creates a number of operational challenges for companies. For example, demand planning, the starting point of manufacturing and logistics processes, must best respond to a series of difficulties created. The constant introduction of new products does not allow sales series to be long enough to provide good forecasts. The reduction of the life cycle of the products also reduces the useful time of the series. In another aspect, the demand series for several SKUs of the same category are not independent, that is, the historical series do not predict interaction in the demand for products. As a result, uncertainties along the chain tend to grow, causing a reduction in forecast accuracy. These uncertainties can cause serious problems with stockouts, high inventory levels and both financial and brand damage.

In this article, aspects such as the portfolio effect and the speed of expansion will be addressed, in addition to a practical approach on how to face the problem. O Demand-driven supply Chain and Big Data will also be addressed given the importance of such topics in the modern scenario of data processing and supply chain management.

2. PORTFOLIO EFFECT AND NEW PRODUCTS

2.1. The Portfolio Effect

As described above, we are experiencing a phase of accelerated growth in the pace of innovation in companies. There has never been so much effort to launch new products and differentiate ourselves, and what is expected for the future is the continuation of this trend, with innovation maintaining its leading position in the generation of sustainable value.

With this amount of products being launched in the market, the consumer has, more and more, different options. In addition, people have more and more information about the products they consume, defining their preferences in a personal way and making their product choices considering various criteria. It is very rare for a customer to be loyal to a brand or product to the point of, in the case of not finding it on a retail shelf, giving up the purchase. In possession of information on what that SKU offers and given the numerous options of similar ones, it is very likely that the customer will opt for a similar product.

Take, for example, the launch of a new canned juice flavor. The launch of this new flavor did not create new space on supermarket shelves, therefore having to compete for space with other products of the same brand, that is, with other juice flavors. This could affect the sales of these other flavors, given the lower exposure of the product. Now let's imagine the sale of a juice with the same flavor, but the SKUs of 350 mL (can) and 600 mL (PET bottle). If a consumer does not find a can of juice for immediate consumption, it is likely that he will consume the 600 mL can instead of the missing can. In these cases, it is noticed that the demands of the different SKUs (whether the same product in different packages or different products of the same family) are not independent, which generates more uncertainties about their prediction. This “cannibalization” between products is known as the Portfolio Effect.

The Portfolio Effect causes even more uncertainty regarding product demand, and in launches this uncertainty is even greater given the lack of historical data on the new product. Therefore, it is important to understand the characteristics of new products in order to define the best way to carry out your demand forecast.

Figure 2 – Launch of similar products may affect the demand for other products in the same family.

Analysis: Wikimedia Commons

2.2. New products

A new product is launched for a number of reasons, including:

• Cost improvement – ​​Product identical to an existing one, but with reduced associated costs, offered to the current market.
• Product improvement – ​​Existing product, but with improvements and superior performance, offered for the current market.
• Line extension – Existing product, but with incremental innovations, offered for the current market.
• New uses – Existing product, but with improvements that allow it to meet new needs, offered for the current market.
• Market extension – Existing product being offered to a different market
• New to the company – Completely new product and market for the company, but not for the competitors.
• New to the world – Radically different product and services being offered to markets served by another type of product or service.

Each of these characteristics can help define the best way to forecast demand, as shown below.

2.2.1. Forecast Statistics for New Products

The choice of forecasting technique depends on several aspects, such as availability of information, time and resources. There are several forecasting techniques, which can be grouped into three categories:

• Judgment: Uses experience and intuition to define the forecast. Examples: Opinion of an executive jury, opinion of the sales force, Scenario Analysis, Delphi Technique, Decision Tree.
• Quantitative Models: Uses mathematical and statistical methods to define the forecast. They can be divided into three large groups: time series (Trend Line Analysis, Moving Average, Exponential Smoothing, Box-Jenkins); casuistic models (Linear Regression, Non-Linear Regression, Logistic Regression); or other methods (Expert Systems, Neural Networks).
• Consumer and Market Research Techniques: Uses information collected in market research. Examples: Concept Testing, Product Usage Testing, Market Testing, Premarket Testing.

Going deeper into forecasting techniques is beyond the scope of this article.

According to the technological characteristics and the target market that the new product presents, a specific strategy for forecasting demand can be chosen, as shown in Figure 3.

Figure 3 – Market vs technology matrix for forecasting new products.

Source: The PDMA Handbook of New Product Development (2012).

When the market and technology are up-to-date, and the innovation brings cost or product improvement, a quantitative analysis of sales is suggested since the product has a similar one on the market. In this case, uncertainties regarding the behavior of demand tend to be smaller.

In the case of a product featuring new technology, but aimed at the same product family as its product market, a product line and life cycle analysis is recommended. This means that a quantitative method can be used based on historical sales data for products in the same line, making an adjustment according to expectations at the beginning of the product's life cycle, given previous experiences with launches. Using judgment methods, such as the Delphi Technique, can be interesting to compare the opinions of different process specialists with the results of quantitative methods.

When the product has current technology, but will be introduced in a new market, a market and customer analysis can be carried out in order to better understand the characteristics of these new consumers and reduce uncertainties about the behavior of future demand.

In the case of a new technology and a new market, uncertainties are even greater and a scenario analysis is suggested. Due to the lack of data, subjective techniques can be used.

It should be noted that, if there are no resource and time limitations, such suggestions can benefit from the combination of different forecasting techniques.

3. RATIONALITY HEURISTICS

The field of behavioral economics had, in 2017, another one of its researchers awarded the prestigious Nobel Prize in Economics. University of Chicago professor Richard H. Thaler presented concepts such as bounded rationality, social preferences and lack of self-control and described how such human traits systematically affect decisions taken, showing why such behaviors have a great impact in the corporate world.

In business environments, where rational decisions are assumed, time and cost restrictions, combined with the growing complexity of the market and its competitive mechanisms, lead to the adoption of a set of simplifying rules, known as decision heuristics, which simplify the process, but lead to systematic errors that are difficult to eliminate.

In the demand planning process, the use of these heuristics can be verified, whether to estimate sales, define the specific impact of a commercial action or measure the success of a product launch. The more complex the environment, the more difficult the information interpretation process is and the more prone the planner will be to unconsciously use some simplifying rule. Let's see the following examples:

3.1. Representativeness Heuristic

The Representativeness Heuristic is the human tendency to draw conclusions from unrepresentative or biased observations. This heuristic is used when trying to answer questions like “what is the probability of object A falling into category B”. This heuristic logic can lead to errors whenever it conflicts with probabilistic logic.

This could occur in the case of launching a new product from a famous brand, for example. Because it is a successful brand, whose products are established in the market, the planner can infer that the launch of the new SKU will consequently be a success. This can lead to overestimations of inventory levels and unwanted costs.

3.2. Availability Heuristics

The Availability Heuristic is concerned with remembering vivid and recent information. There is a tendency to consider a high probability of occurrence of a given event because the same event has occurred recently. For example, when there is a plane or car accident and people start to avoid taking such means of transport because they think that the probability of their occurrence increased after the event.

It is not always easy to distinguish what really impacts the probability of an event occurring. Understanding the availability heuristic is essential in exceptional situations in demand planning, such as promotions and sporadic events.

In the case of a new product launch, it is common and well-regarded to recall previous similar experiences to serve as a starting point for demand planning. However, its use, without considering changes in the commercial context, new market and economic conjunctures, changes in competition or in consumers' purchasing power, can lead to relevant errors. Relying only on memories is a shortcut that can lead to great losses.

3.3. Other Heuristics

Other heuristics, such as Anchor and Adjustment, Backward Forecasting and the Confirmation Trap, also play an important role in the demand planning process. In addition to having a structured process, based on the use of statistical models for forecasting sales and reconciling commercial and operational plans, good planning depends on planners who know these heuristics and are able to recognize their impacts and characteristics, in order to avoid them.

4. DEMAND-DRIVEN AND BIG DATA

It is expected that the best answer to this dilemma between the use of sales forecasts and the constant launching of new products is in the supply chain focused on the real demand, or the Demand-driven Supply Chain. The less the chain relies on historical sales forecasts and getting, in a timely manner, real demand information to trigger the other links in the supply chain, the better for forecasting accuracy and the less likely it is to incur high costs associated with disruptions and excess stocks.

In addition, the advent of Big Data technologies will allow the processing of information from various sources regarding consumer preferences and choices. It is possible that, in the near future, market research, which today is costly and time-consuming, will take place in real time and continuously, to provide important information about demand. Other technologies such as Artificial Intelligence, 3D Printing, Drones and the Internet of Things will allow chains to be agile and flexible to respond to this new paradigm of meeting demand.

5. CONCLUSION

As innovations and launches of new products and services will continue to grow and be important for companies, it is up to demand planners to deal in the best way with available techniques to act in forecasting demands in a context of high uncertainty. Collaboration is essential for information to flow quickly between departments and links in the chain. The use of certain forecasting techniques must be done according to criteria such as product and market characteristics and available time/information. Another important care is to avoid the use of heuristics and biases for decision making, which can lead to lower than expected results.

With the advent of new technologies that allow greater collaboration between the links in the chain and the collection/processing of more information in real time, it is expected that these uncertainties will be mitigated and a greater alignment between forecasts and actual demand will occur.

SUGGESTED BIBLIOGRAPHY

• KAHN, Kenneth B. The PDMA handbook of new product development. John Wiley & Sons, 2012.
• STARK, John. Product lifecycle management. In: Product Lifecycle Management (Volume 1). Springer International Publishing, 2015. p. 1-29.
• WANKE, Peter; JULIANELLI, Leonardo. Sales Forecast: Organizational Processes & Quantitative And Qualitative Methods. Publisher Atlas SA, 2000.