IP&O solutions help you see the big picture while driving results.
If you have worked for even a little while in inventory management, you’ve developed an appreciation for the complexity of the job. You’ve also probably realized a few important facts along the way:
Sometimes, it seems like a job in supply chain would be easy if not for the constant randomness: Customer demand is random. Supplier lead times are random. Raw material prices are random. The global value of money is random. If you lived in a world without randomness, you could figure out exactly what to do and when to do it, in order to help your company make the greatest profit.
In the real world, you need demand forecasting and inventory modeling to help you harness that randomness and transform it into quantifiable results. These methods are based on math and statistics, so it is helpful to let software deal with the technical details while you steer the ship. Models can tell you, “Here are the risks and rewards if you do such-and-such, and here is another version if you do this instead.”
There are models and there are models. Consider forecasting—it is tempting, but ultimately foolish, to rely on forecasting models that promise, “Here is what the exact demand for Product ABC will be six months from now.” That’s a fairytale.
Suppose you have a perfectly accurate forecast of the most likely demand for some item in the next planning period. That’s helpful, but also grossly insufficient; the actual demand is almost guaranteed to be some other number. You need a more comprehensive forecast that will predict the entire probability distribution of demand, not just the most likely value. When you see the bigger picture, you can assess the risks created by variability in demand and adjust to hedge against those risks.
To beat the odds, you first need to know the odds. Probabilistic Forecasting is the modern successor of traditional “one-number” forecasting methods. It outputs a full range of possible item demand levels, using probability theory to attach odds to the possibilities.
The forecasts are not the end of the planning process, but rather the beginning. To be effective, they must be fed into models of stock levels. The basics of modeling inventory require only simple arithmetic: for instance, the number on hand at the end of today will be what you had at the end of yesterday minus any shipments made today, plus any replenishments that arrive today. But since today’s demand and replenishment numbers will be random, there are many ways things could play out.
Modern software can generate thousands of scenarios to expose the full range of possibilities. This is called inventory simulation. As a general rule, really complicated systems are too much for traditional paper-and-pencil math, so these simulations are the only way forward.
The way the scenarios develop depends not only on the randomness in demand and lead time, which you can’t control, but also on your choices of things you can control—namely, inventory policy and parameters.
For instance, one inventory policy (among others) that you may choose asks you to specify, for each inventory item, the values of two inventory parameters: a Reorder Point and an Order Quantity. If you set the Reorder Point at 10 units and the Order Quantity at 25 units , you would generate a replenishment order for 25 more units when stock on hand drops to or below 10. The policy must also specify what happens when there is a stockout. Do you lose the sale, or will the customer accept a backorder?
Only when you have modeled demand and lead time uncertainty, while also specifying an inventory policy and parameter values, can you generate meaningful simulations. At that point, the software can tell you, “If you run your inventory this way, here’s a full picture of all the ways this could play out.” Some of those ways would be good for your business and others would not, so the analysis lets you assess the risks and rewards of your decisions—before you commit to them.
The next step in the decision chain is to decide whether you like what the simulations say will happen. Now, it’s not a customary thing for most of us to be confronted with a large amount of numbers and decide immediately whether they make us happy. Software simplifies the problem by calculating summary performance metrics from the simulation results. These metrics put the consequences of your decisions for managing an item in dollar terms within three cost categories: holding costs, ordering costs, and shortage costs.
Your accountants should be able to tell you what it costs to hold one unit of any item in stock for one day. Likewise, they can estimate the paperwork costs of placing one replenishment order. A bit trickier is the estimation of the dollar cost of not having a unit in stock when a customer requests it—there is obviously a loss of revenue, and sometimes there are penalties incurred, and there is always some loss of intangible “goodwill.”
Some companies monetarize all three cost components. Others prefer to treat stockouts differently and express them in terms of the percentage of units demanded that are not immediately available for sale. The two metrics that are commonly used to measure stockouts in somewhat different ways are service level and fill rate.
All this analysis comes together when the software exposes the links between the choices you can make and their consequences. Since choices about inventory control parameters can decrease some cost components but increase others, you need to consider how things net out.
For instance, suppose you were to increase the reorder point for an item. This will have both good and bad cost consequences. The average on-hand inventory will increase, so holding costs will rise. On the other hand, the time between stockouts will increase, so the cost of reordering will fall. The effect on shortage costs is less easy to anticipate, but the simulations will show how it can all play out.
It is worth noting that the simulations can also help you think creatively about potential changes in factors that are less directly under your control. Consider supplier lead time: you may feel stuck with accepting whatever lead times your supplier chooses to offer (e.g., “We’ll try very hard to respond within one week”). If this results in erratic performance, would it be worth it to you to reward your supplier for greater consistency?
Re-simulating your operations with less variability in lead times will reveal the benefits of more consistency in replenishment, helping you think about changing the terms of your relationship with that supplier. In this way, the simulations can open new possibilities for improvement.
Software can take the analysis of tradeoffs to various levels. At the simplest level, it can give you an interactive view of the consequences of particular decisions: If I increase my reorder point from 10 units to 15, what will happen to my holding, replenishment, and shortage costs?
At the next level, software can show you all the design choices that meet some specified criteria, such as, “holding costs plus ordering costs below $X and service level at least 85%.” This reduces an infinite number of choices to a finite set, which you could choose among based on other criteria such as how much work would be required to implement each one.
At the highest level, if you choose to express shortage costs in dollar terms, you can have the software automatically sort through a huge number of system designs and identify, for each item, the values of reorder point and order quantity that would result in the lowest total cost of managing that item. This is inventory optimization
Precision and accuracy are critical for success across the supply chain. The right forecasting and modeling tools can help you make smart decisions for the long-term growth of your company.
Learn how Epicor solutions make every day easier for your business.
