The Math - Introduction

In the initial article that frames this ongoing story, I mentioned the ‘COVID hangover’ that has affected many aspects of supply chain planning as a result of the past 5 years.? The volatility that we felt across supply, demand, and distribution was so extreme that it challenged our fundamental understanding of what supply chain planning, and the key inputs, really mean.? The first client of my career was a grocery store chain and at the time I didn’t realize how helpful it was to be able to connect the complexity and scale of a global supply chain to our own consumer behaviors. ?To really understand a supply chain, you must begin with an understanding of your end consumer, their behavior, and what drives volatility in this behavior. ?Most B2B-focused businesses do not have this natural understanding of the customer’s needs, but it is a critical step in the process.

So, think about your own home and how you plan the supply of food.? Whether you rarely cook a full meal at home or your family of 6 relies on $1000 weekly Costco runs, there are many items that you intuitively plan and procure on a regular basis. Eggs have been a classic headline-news example in recent years as an always-available, cheap commodity that suddenly was neither. First, the demand for eggs is very stable and predictable… typically.? Most families know the rate of eggs that they consume on a weekly basis, where they like to get them from, and what type of eggs they prefer. ?This stable consumer pattern means that even at a large scale, short-term demand planning can be a simple model that incorporates some level of seasonality and some preparation for disruptions.? Longer-term planning accounts for the growth of a population served and major changes in competition.? A grocery store’s demand for eggs is an item that we can accurately predict on a horizon that only needs to be relatively short. ?However, we can all think back to the shortages, stockpiling, and crazy habits that began in our own homes in 2020.? Mom or Dad may have picked up breakfast on a commute, or kids ate their breakfast meal at school.? Then in 2020, this all changed. Demand went well outside of what were normal planning parameters.? In a week’s time, a school cafeteria’s demand for pallets of eggs went to 0, and every home across the globe began cooking breakfast in their kitchens. These fairly fixed supply chains couldn’t redirect enough eggs from the restaurant or institutional channels to the consumer channels.? Because of this, there were records for spoilage and waste while consumers saw bare shelves.? In the coming months, hatcheries were shut down at alarming rates, shortages continued, and prices went through the roof. This extreme example will serve to shed light on how our changing behaviors impact supply chain planning efforts and the inputs we must model.?

Known Inputs

In our work in this space, we talk a lot about information that can be defined as known vs. unknown variables that we need to model with flexibility and sensitivity. For the known metrics, we just need to represent these in the math in a way that is consistent and trusted, think about sales price and cost, weights, and dimensions. How much does a case of eggs weigh? In some instances that data may need to be accurate because you may fill a truck from that vendor by volume well before weight limits are met. However, when demand shifts, your old normal order from that supplier is irrelevant, and all of a sudden we need to build out a full truck from a supplier that has a totally new mix. ?Identifying these data weaknesses is a significant step in the process when you think about shifting from traditional buying patterns to an optimized and more flexible flow. However, this first step is a data challenge, not a modeling challenge.? This serves as the first step in the process and is often a major hurdle to doing systematic analysis on a large scale: can we trust our data?

Unknown Variables

The second type of information is more interesting and where we will focus here. ?These are the characteristics of the model that are explicitly unknown and must be modeled to predict a relevant range of outcomes.? In this case, our outcome is a target inventory level for each SKU in the system.? For this explanation, we will think of a distributor. One that exclusively buys and sells finished goods.? This is a fair foundation to discuss the fundamentals, with the complexities of Bills of Material, Kitting, value-added services, or engineered manufacturing all introducing further complexities.? At the simplest level, and whether it is strictly managed in this way or not, there are three main reasons for holding inventory on your books within a system: in-transit, cycle stock, and safety stock. In-transit is largely a financial consideration and a contracting challenge.? How long does inventory sit on your books? What are the triggers for payment to your suppliers?? Cycle stock is largely driven by the replenishment frequency and an economic order quantity, which we will discuss later. ?The largest reason for carrying excess inventory, and the value that we can most quickly have an impact on, is the Safety Stock or Buffer Stock that is held, purposefully, in the system. ?Safety stock is held for two main reasons: demand being higher than expected and supply being lower than expected.? There are lots of calculations that have been published over the years, but the most common formula for statistical safety stock is shown below, which attempts to model the variability of supply and demand as random, normally distributed, independent variables.? Each of these three assumptions is easy to challenge and disprove the underlying theory, but it is a great place to start and does help to define the need.

In the next article we will go into each aspect of the supply chain math, considerations of each variable, and what this could mean to your business today.