Want to Save Millions? Watch Your Milliliters.

In today’s fast-paced world, corporations cannot stress enough on quality. With increased globalization and the advent of the Internet and social media, people not only have more choices but they are also aware of those choices.

With the clutter of marketing messages consumers are exposed to on a daily basis, brand loyalty is becoming harder to build and maintain. If a customer today has a bad experience with a particular product, he or she is not only likely to switch to a competing brand, but can also be expected to share that unsatisfactory experience with friends, family and others on social networking sites.

Soft Drink Manufacturing Facility

Operations management plays a significant role in the maintenance of quality in an organization’s products and processes. One of the most popular methods for quality control is Statistical Process Control (SPC), an analytical decision-making tool that facilitates the monitoring and control of processes. It allows one to examine a process in order to detect any variation in it that might require correction.

For instance, at a soft drink manufacturing plant, SPC may be used in the production process in which the finished product is filled into PET bottles. The cola filled into a 500mL is hardly ever exactly 500 milliliters; it could be 500.04mL, 499.98mL, 499.93mL, etc. Statistical process control will use a sample of bottles filled at a particular plant to determine the variation in the average volume filled.

I came across a practical application of SPC during an internship at a global manufacturer of consumer goods. A Statistical Process Control analysis at a shampoo-manufacturing facility revealed that the liquid volume filled in 400mL-shampoo bottles was consistently ranging between 400.4mL and 400.9mL. Although this is even less than half a milliliter, a large company could have suffered significant unnecessary costs if consumers were constantly given more than 400mL shampoo in the long run.

Stages of Statistical Process Control

The management suspected that the volume irregularity was not due to a natural or common cause. Control charts constructed for the bottle-filling process confirmed this notion. The variation lay outside the control limits and was therefore due to an assignable cause not part of the original process design. It turned out that one of the levers in the filling machinery was not functioning correctly and allowed more liquid to enter the shampoo bottles than it was designed to fill. The SPC analysis consequently allowed us to identify this problem fairly early and re-calibrate the equipment before much money was lost.

An article by Manus Rungtusanatham in the Journal of Operations Management states that the benefits of SPC are much more than just improved quality and cost cutting. Research has shown that the implementation of statistical process control in production environments works to motivate process operators. As these front-line workers become more satisfied with their jobs, they are more motivated towards continuous improvement and high quality.

With all its advantages, SPC does have some limitations. When performed regularly, continuous inspection can be quite expensive. While the cost may be justified for a large manufacturer such as P&G, is Statistical Process Control as relevant for smaller companies too?

Why Guess When You Can Forecast?

In the last class we studied the concept of demand forecasting and its importance in the business environment, especially in the 21st century. This topic was particularly interesting to me since I have recently experienced problems related to inaccurate forecasting at work.

I work for a supply and distribution company that deals with the fast moving consumer goods (FMCG) industry. While previously we had only undertaken the distribution of home and personal care products, recently our company decided to add several food product brands to our portfolio. This was a strategic move since we are experienced in dealing with various Middle Eastern retail establishments. However, the mistake our team made was to purchase the product inventory from manufacturing companies without accurately forecasting the demand for those products.

The Role of Forecasting in Demand & Supply Planning

The result? We ended up with far more inventory than we could sell. Food products are perishable; their expiration deadlines are much shorter than for other consumer goods. As those expiry dates approached, a considerable percentage of the inventory we had bought was wasted in our own warehouse. Needless to say, the company suffered some heavy losses.

Few forecasts are absolutely accurate since the future is inherently uncertain. To add fat to the fire, many companies still use informal forecasting methods such as educated guesses by top management and intuition or “gut feeling”. Some use quantitative methods such as historical sales trends and adjust it according to the forecasting officer’s own personal experience or opinion. When forecasting methods are based on such subjectivity, how accurate can they really be?

Demand Forecasting

Concerned about the accuracy of my company’s forecasting methods, I researched various forms of quantitative and qualitative forecasting. I came across an article by Kesten Green and Scott Armstrong of Wharton. The article proposed that only structured forecasting methods should be used and more qualitative techniques such as focus groups, unstructured meetings and intuition should be avoided. Even when some judgment must be used (possibly due to the lack of sufficient data), Armstrong and Green (2005) recommended that forecasts should be based on more structured procedures such as the Delphi method or structured analogies. The structured analogies method involves using results from similar situations from the past to predict the outcome of the current situation using a structured, formal process.

Even when accurate forecasts are made, it is not always easy to implement them in business decisions. When research results reveal figures contradictory to what the top management expected, they may even be ignored. To increase the acceptance of forecasts, decision-making managers can be asked to agree on what methods should be used before any forecast results are presented. The scenario approach can also be used; decision-makers can be asked what steps they will take in different possible future situations, before revealing forecast results. For instance, the managers at my company could have been asked what they plan to do if forecasts reveal that demand is considerably lower than our inventory of food products, and what if demand exceeds supply. This way, they are more likely to act on the results of the forecast.

Whatever the method used, companies should focus on maximizing the accuracy of their forecasts. In today’s fast-paced world where competition is ready to grab your market share at the slightest miscalculation, I feel forecasting is critical to business success.