A Quick Method for Assessing Inventory Health Posted
Last week we discussed a Gartner framework for gauging inventory health, but noted that the Inventory Quality Ratio (IQR) could take a while to implement (Click here to read last week’s blog). This week we look for a relatively quick way to assess inventory health, so we turned to Gerrott Faulkingham, solutions architect at ToolsGroup. "It all starts with a demand-driven approach," says Gerrott. "Even though we are talking about inventory, we first have to understand customer buying behavior—and how that then translates into inventory requirements."
As a reminder, the key to measuring inventory health is to assess holdings relative to expected future customer demand according to service level expectations. You want to measure how much of your inventory is “in the zone” where the level is between defined minimum and maximum levels based on stock coverage of projected forward-looking demand. The typical steps in the process include:
- Base case analysis and service level simulation
- Defined scenarios and sensitivity analyses
- Quantified value proposition and go forward plan
Base case analysis and service level simulation - The assessment starts with the presumption that healthy inventory supports a target service level for a class of products. This is different than traditional ABC inventory classification (to see why, see this blog). The goal is not just to understand how much inventory is available, but what service levels are being achieved without the aid of expediting and other forms of manual intervention.
Faulkingham explains, "We take multiple pictures—similar to a ‘Monte Carlo’ simulation of many possible outcomes—to benchmark the service level percentage at which a company is currently servicing its end customers. Often, it turns out that firms are servicing their customers at, say, 92%, when they think they are closer to 95-96%. The gap between these two figures is filled by the heroics of the planning team putting in extra hours to chase down and expedite orders."
Defined scenarios and sensitivity analyses - The next step is to determine optimal stock holding targets for each location based on current overall service objectives. This means evaluating inventory vs. service level tradeoffs at various service level thresholds. Typically companies assessing inventory health have one of two objectives—maintain present service levels but reduce inventory, or increase current service levels with the same amount of inventory. For example, one multinational wanted to maintain service levels but reduce stock—spend less, yet achieve the same aggressive service levels it was hitting today, on the order of 99.2%.
This requires software that models the demand and calculates the probabilities, or ranges, of demand that can occur. Demand modeling understands inherent uncertainty and can develop probabilistic models of demand patterns based on variables such as order frequency.
"The crux of the matter is understanding the differences within the same forecast—the varying levels of demand certainty for each item within that plan," Faulkingham says. “This translates directly into inventory—the more certain you are about your forecast, the tighter you can keep the controls around how much inventory you need to carry to meet that particular forecast."
For example, say two items sell 100 units per month—one is 100 orders for one unit, and the other is a single order for 100 units. These two demand profiles have entirely different levels of certainty. With the one order for 100 units, if you lose one order, you lose your entire demand for the month. With the other item there are 100 different orders, if you lose one you only lose one unit. You are more certain about the forecast when demand trickles in vs. a single bulk order. So order frequency and order size impact demand uncertainty (shown as the vertical blue forecast bars in the image below) which then directly impacts inventory required.
Quantified value proposition and go forward plan - The final step of a quantified value proposition. For example, one modeling exercise determined that the firm was overstocking its fastest-moving items. Faulkingham says, “A global optimization of almost 650 SKUs demonstrated how $1.5 million of current stock could be reduced by over 35% to under $1 million in working capital investment—without impacting service levels.”
The go forward plan is usually relatively easy from a calculation standpoint. The challenge is to present the alternatives in a meaningful way so that the management team understands and can make an intelligent choice based on inventory tradeoffs.
In addition, the benefits need to be realistic and achievable, so the modeling needs to reflect lifelike conditions like lot sizes and lead times. In other words, the model used to optimize inventory needs to be “robust”, not “brittle”. In the end, Faulkingham reminds us, “You have to be able to get what you plan.”