Supply-chain leaders everywhere are redefining what it means to achieve "resilience" — and it's growing beyond stockpiling and risk response. In the wake of the COVID-19 pandemic and other disruptions, resilient supply chains actually thrive in chaos, and crucially, offer better customer service than before.

To make the concept of resilience tangible, let’s connect it to something that we as supply-chain professionals all understand: inventory. Inventory that’s optimized to meet customer service levels isn’t about raising safety stocks, but quite the opposite. Some planners during the pandemic have increased overall service levels by 3-5% while decreasing overall inventory levels by 10-30%. Like so much supply-chain theory, this feels counterintuitive. Many planners naturally assume that inventory levels must increase to increase service levels.

Order-Line Fulfillment Rate 

To unpack this contradiction, we need to distinguish between the ability to serve aggregate demand and the ability to fulfill customer orders. Here I’ll repeat the mantra I say to all my customers: order fill — or order line fill rate (OLFR) — is the metric that matters when it comes to hitting target service levels. Customers only care about whether they received their orders on time, not whether your aggregate demand forecast was accurate.

But isn’t an accurate forecast the main indicator of a supply chain’s ability to fulfill orders? Therein lies the misunderstanding. High forecast accuracy might scan well on an executive dashboard, but it says nothing about the order fulfillment rate. Why? Because forecast accuracy, as measured by most planning tools, only relates to total demand, not the ability to fulfill individual orders.

Here’s a simple example. Let’s say you get 10 orders in a month for a single LED lamp to be shipped to 10 different parts of the country on different days. Of course, fulfilling those orders requires a totally different plan than an order for 10 lamps to be shipped to a single location. However aggregate forecasts make no distinction between the two. This is exactly how planners and warehouse workers arrive at the 99% service level. As long as there is only one of each item in the warehouse, 100% of the products are "available". But it only takes a single order for 10 LED lamps to throw the system into disarray. Not very resilient!

I can’t emphasize strongly enough how important it is during our current era of high uncertainty to set inventory policies that ensure high OLFRs. According to the McKinsey study "Adapting customer experience in the time of coronavirus", during the last recession those companies that offered the best customer experience achieved three times higher returns than their competitors.

So how do we plan for high OLFRs? Spoiler alert: it’s not ABC segmentation. This approach, developed in the 1960s, doesn’t work for today's sales complexity, volume, and diversity of SKU portfolios. Planners using ABC end up applying the same inventory policy to vastly different scenarios like those two LED lamp scenarios above and end up with unacceptable surpluses and shortages. ABC segmentation was suitable in its day for managing stock levels, but it was never for optimization.

Inventory Mix Optimization

Unlike ABC, which segments SKUs into three categories based on "business value", inventory mix optimization establishes a blanket inventory policy for each SKU. Inventory mix optimization segments items by "service class" and then assigns a unique service level and inventory policy to each individual SKU. These more customer-focused categories should be relevant to sales and marketing — for example: "accessories", "high margin items", "private label", "high value brands" and "critical spares". By grouping items in ways that are meaningful to customers, you set inventory policies for each that correspond to service expectations.

Inventory mix optimization depends on using advanced planning software to apply "inventory-to-service" curves to optimize service levels and safety stock for each SKU location. The stock-to-service curve shows the relationship between the desired service level (OLFR) and the average required stock level, taking into account the required control levels (safety stock, reorder stock). A basic inventory mix optimization principle is to assign a lower service level to certain service classes — such as long-tail items that are rarely sold and that customers do not urgently need.

David Lubinski Ltd., the sole importer for Citroën and Peugeot automobiles in Israel, switched to an intelligent planning system to optimize its inventory of about 20,000 parts, 75% of which are slow-moving items. Though this established, family-run company had been profitable with its ABC inventory planning, it suspected that resilience and service levels could be improved.

To test the switch from ABC segmentation and spreadsheets to a new system focused on inventory mix optimization, Lubinski carried out a side-by-side comparison of both approaches. The results were conclusive. Lubinski cut its inventory by 25%, reduced air-freighted rush orders by a third, and saved considerably on scrap and obsolescence. It achieved all this while maintaining well above average overall service levels of 95-96% and providing individual service levels tailored to the needs of each product. This enabled the company to return €1.5 million to the bottom line through inventory savings alone.

Planner productivity also increased. Lubinski's new system is now almost completely automated. Today, one person spends less than a weekday managing the process, so the planning team can now focus on more important, customer-facing work. The only thing Lubinski's spare parts planner regretted on the eve of his retirement was that they had not introduced the new system sooner.

David Barton is general manager at ToolsGroup NA.