Multi-Variable Mayhem

The trouble with managing inventory is the many variables that affect stock levels. This balancing act is hard to do because there are so many different variables that influence decisions:

• Demand Fluctuations
• Supply Fluctuations
• Inventory Policy
• Working Capital
• Manager’s Intelligence
• Monitoring & Management Systems

We tend to think of the process in simple terms, making sure to keep everything in stock but not holding too much stock. That is the goal — a Goldilocks story of not too cold and not too hot. The problem is far from simple. Consider the causal diagram to below, showing the core system, Inventory Level, and the six influencing systems that affect the central system.

External and internal forces influence Demand Fluctuations and Supply Fluctuations. The internal decisions that influence demand include advertising, promotions, placement, and available substitutions. Companies can drive demand for a product through promotional pricing and advertising. Internal decisions about suppliers, like sole-sourcing, importing, and direct vs. wholesale channel — affect supply performance. The speed of demand can drive the system faster, and a lag in supply can slow the system down. These two variables are the velocity systems that drive the speed of decision-making. That is why we show the snowball descending the hill and the stick figure pushing it back up.

The other four influencing systems rely on internal decisions and on the decision-making ability of internal managers. Each of these remaining four systems is a balancing system that seeks an internal balance in order to place a braking action on the inventory level in an attempt to balance across all the influences. We show a balance in each of these influencing systems because each subsystem attempts to balance internally first, influenced by the needs of the central system, and in turn influencing the central system.

The balancing systems balance other influencing systems. Our model shows Manager Intelligence opposite of Systems and Data, and Inventory Policy opposite of Working Capital. A very smart and resourceful manager can operate with limited systems and data, while an inexperienced manager relies much more on the systems and data to do the job right. Limits on Working Capital capacity require more refined Inventory Management Policy.

These are not the only factors that influence the not too much/not too little balance of inventory management; the other factors influence one or more of the six main influencing systems. A supplier’s variable performance is but one factor driving Supply Fluctuations. Advertising is a factor that drives demand. In fact, a competitor’s promotional advertising campaign can drive demand up, if the competitor’s price is high or they have lower quality.

This is a multivariable problem. People have a hard time with multivariable problems. We look for simple rules of thumb, and then add safety factors to our calculations to ensure that our stocking levels remain adequate. These safety factors tend to add more inventory to your plan, increasing the pressure on limited resources like storage capacity and working capital.

Computational Horsepower — Cognitive Hangover

Just a few decades ago, few inventory managers could contend with the computational challenges of the multivariable analysis. The limitations of available systems and data hindered inventory planners. The planner of the 1980s and 1990s worked in delay loops, following the trends and adjusting safety stock values as they could. Real-time systems just did not exist except in the most forward-thinking, well-capitalized companies. There were few real forecasting systems; most companies depended on past sales history mixed with the planner’s experience. Inventory policy depended on a mixture of deeper stock, driven by higher service levels and greater levels of committed working capital.

By the mid 1990s some forward-thinking planners had started to use desktop personal computers and spreadsheet software, like Lotus 123, to develop safety-stock planning models. These planners, with support from their MIS departments to get data downloads, started to develop analysis models in an effort to tighten up the amount of safety stock they used to avoid the OSWO (Oh Shoot, We’re Out) situation. As smart as these intrepid planners were, they still worked in relative darkness, since few had access to supplier performance data.

Major companies with the capital to make investments built or acquired systems that integrated warehouse and inventory systems. Inventory accuracy in warehouses improved, reducing stock-outs created by systems integrity issues. Demand forecasting systems evolved to include external demand signals, and planning adjustments based on human intervention.

Each increase in computational horsepower improved the ability of organizations to manage inventory. Additional horsepower was not the only boost; interconnection of the systems of different companies, via EDI, helped increase the velocity of data transfer between companies. Software increased in capability, while interconnections increased the complexity of the systems. Companies had to reduce costs, justifying the deep investment in systems with a combination of reduced headcount costs and increased working capital leverage.

Faster and more powerful systems and software did improve inventory performance … for a while. Over time, the inventory planners who built their own analysis models in Lotus or Excel, retired from the stage. Their replacements, despite having been exposed to safety stock and lead-time requirement calculations in college, had lost sight of the factors that drove the calculations in those systems. The increased computational horsepower turned the inventory systems into black box systems — users fed data into the system and expected perfect orders to come out.

Nothing could be farther from the truth. While these systems could hammer through the millions of calculations needed to manage Demand Forecasting, determine SKU level stocking levels, order point calculations, and create Purchase Orders to issue to vendors via EDI, the systems were not dynamically reactive. Variations in vendor performance, never collected, proved to create just as many stocking problems. Late deliveries and poor fill-rate performance created stock-outs. The real root cause eluded many inventory planners, partly because they underestimated the impact of inbound variance.

These planners underestimated the impact because they lacked an understanding of the safety stock calculations. Increased computational horsepower created a lot of cognitive hangovers in the inventory planning departments of many companies. I’d go as far as to say that most inventory planners working in either the retail or manufacturing worlds could not calculate safety stock by hand (or with a spreadsheet) without finding reference materials with the formulas.

How Replenishment Squeeze Plays Happen, and the Defense Against Them

Deconstructing Safety Stock