I shouldn’t have my blood pressure taken after reading some of the widely distributed business periodicals. I just finished reading a newsletter where the author states that distributors should abandon forecasting future demand of products and merely continue to replenish inventory by replacing quantities based on what customers have just ordered. For example, when a customer buys 10 pieces of a product, you immediately replace those 10 pieces on your shelf. He draws a comparison to a vending machine that is filled up based on recent purchases, not expectations of what customers will want in the future. While this “pull” system may sound great in theory, most distributors, retailers, and manufacturers will have a hard time successfully implementing it in their real-world environment. Let’s look at several of the author’s assumptions [and my rebuttals]:

  • What you sold in the immediate past is a good prediction of what you will sell in the immediate future.
    [Jon]: The author is totally ignoring seasonality and other trends in demand. What if a product loses popularity? Won’t you be stuck with a lot of dead inventory because it was selling well a while ago?
  • You have the manpower to continually receive small shipments (say a one- to three-day supply) of most if not all of your inventory items.
    [Jon]: How much labor can you afford in your warehouse?
  • Your suppliers are willing to make and send you small quantities of all of their products.
    [Jon]: The author forgets about higher freight rates for small shipments; the time necessary for manufacturers to coordinate and set up the manpower, equipment, and raw material necessary to produce their products; as well as other realities of the real world.

The author compares his “revolutionary” “pull” system to “just in time” (also known as “JIT” or Kanban) systems used by many manufacturers. For example, car manufacturers will take delivery of only the parts they need for immediate consumption on the production line – say, one day’s or even one hour’s supply of headlights. But he forgets that the car manufacturers had to start with a forecast of the number of cars that would be sold in the future in order to coordinate the machinery, manpower, and supply of incoming parts to make the cars. Does he really believe that when I bought my last car the dealer called the factory and told them to immediately make another car just like mine and deliver it to the lot tomorrow?

We work in the real world. All of our concepts, tools, and theories are developed working with our clients to better achieve the goal of effective inventory management. A critical element of this goal is as accurate a forecast of future demand as possible. This forecast is necessary:

  • To be sure you issue replenishment orders at the right time. If you predict you will sell two pieces per day and an item has a seven-day lead time, you had better reorder the product when there are 14 pieces left in inventory (2 pieces per day * seven days = 14 pieces).
  • To be sure you have enough stock to last between replenishment shipments. If your vendor requires you to order a truckload of his products and you sell a truckload of his items every thirty days, you probably want to bring in a replenishment shipment about every 30 days. You’d better order no less than what you predict you will sell of each item (i.e., your forecast) in a thirty-day period starting at a time equal to today’s date plus the lead time – that is, a time period starting at when you would receive the shipment if you ordered it today.

Forecasts are predictions of what you will sell or use in the future. Predictions are never 100% accurate. Often they are not even close. But the answer is not to ignore reality and adopt a system based on the belief that it is practical to continually order and receive small quantities of every product and that your customers will continue to buy exactly what they purchased in the immediate past.

In a way it is a good thing that these beliefs aren’t true in the real world – because if they were true, it would take no skill to be successful in business.

Next month, in Part Two, we’ll examine approaches to evaluate the merit of business articles we read. Over the next several months, we will look at practical ways you can improve your forecast accuracy. In the meantime, remain skeptical! Don’t believe everything you read.