Mastering Lumpy Demand - Part 5

Most supply chain practitioners have recognized the inadequacies in planning for long-tail demand using deterministic solutions. So, they have attempted to address the problem from the supply side through tactical measures. This includes adopting philosophies like the Theory of Constraints (ToC) and developing consulting practices around supply-driven software. Their approach often involves supply chain segmentation, resulting in simplistic safety stock strategies and KANBAN-based replenishment rules. Due to inadequate technology for addressing demand-side issues, they focus on the supply side instead.

Additionally, they implement risk pooling and position decoupling points further upstream in the supply chain—either at the component/assembly level in a manufacturing network or at the regional DC level in a distribution network. These tactics again rely on simplistic assumptions about demand and ultimately compromise service levels. They may also attempt to control sales under the guise of supporting a holistic fulfillment strategy, using pricing incentives to forward-pull demand signals and allowing longer fulfillment time periods to manage inventory. However, these tactical measures create additional bullwhip in the supply chain, exacerbating the problem they aim to solve.

Another Shortcut: The Two Model Approach

Several software vendors have realized the serious inadequacy of using the normal demand distribution for modeling slow-moving item behavior. That’s why you would hear software companies publicizing the launch of new algorithms to tackle intermittent demand. For those who have tried to solve the problem, the most common remedy has been to create a second model, based on one of the many derivations of Croston’s method, explicitly dedicated to “slow movers”.

Croston’s method was originally created in the 1970’s and is based more on empirical considerations than on a sound scientific basis. It has proven useful to avoid crazy forecast behaviors of intermittent demand items, but nothing more than that. It doesn’t provide any reliable way to drive inventory, especially set safety stocks, and target a specific service level.

Finally, the two model approach requires a-prior classification of the SKU’s, separating the “normal” ones from the “lumpy” ones. Products are split into two separate categories, making use of two entirely different modeling techniques for the two categories. All observed real world phenomena present an infinite variety of different behaviors, not just two extremes. Inventory modeling is no exception to this rule.

The futility of forecasting using time series methods and then setting broad-brush ABC inventory norms to tackle lumpy demand has been detailed in previous editions. To address lumpy demand, we need a dual-engine, demand-driven inventory models that’s service-driven. They work in tandem and in an intertwined manner. When one lags, the other should automatically step in to complement within the laid-out business, financial, and operational constraints.

Recognizing the futility of forecast accuracy for planning long-tail items, the real question should be: Can we truly understand the demand implications on the inventory behavior for a large number of long-tail items well enough to achieve “service level excellence”? With the right technology, the answer is yes.

To achieve this, you must master the shape of the demand distribution and the right-hand tail. Avoid modeling shortcuts and approximations that can prevent you from hitting your target. Achieving high service levels in a lumpy demand world requires the following capabilities:

• A very reliable demand modeling technology that automatically adjusts all relevant statistical parameters, in a seamless way across a wide range of SKU behaviors, understanding all elements of demand uncertainty and considering not only forecast errors, but also the “intrinsic demand variability” caused by order-line frequency and order-line size distribution.
• Advanced inventory modeling that, by eliminating many of the gross approximations of traditional inventory management theory, can provide very reliable descriptions of statistical inventory behavior.
• Logic for demand signal propagation that combines the superior capabilities of both demand modeling and inventory modeling, to model the impact of replenishment parameters and constraints at each echelon of the supply chain. We have detailed "multi-echelon replenishment" capability in another edition.
• A solution that assures very good adherence between plans and actual performance, reducing the need for emergency interventions, and thus allowing disciplined inventory and service planning processes. Also read the edition titled: Following the Money: From S&OP to S&OE in Supply Chain Investments.

Summary

We have observed a series of issues that make lumpy demand and the long tail environment challenging. First, lumpy demand exhibits high variability and highly skewed distributions. As a result, classic demand and inventory models perform poorly in this context, leading to performance gaps that necessitate manual interventions and further contribute to the bullwhip effect. Consequently, inventory becomes misaligned, and the company's business goals are unmet.

The solution is to master complete demand behavior; variability, frequency, and order line size distributions. Mastering lumpy demand involves accurately and reliably modeling both head and tail demand seamlessly. This must be done without taking convenient shortcuts, which will simplify the problem and lead to inaccurate models and erroneous decisions. By adopting a “no-shortcuts” approach, the primary benefit will be increased precision (both demand and inventory) and user confidence. This will minimize manual overrides and eliminate the bullwhip effect in the supply chain. By mastering the full probability distributions of demand and inventory across various behaviors, you can achieve unprecedented efficiency and service level excellence in a world full of lumpy demand.