I interviewed David McPhetrige who discussed 'Why I am obsessed with Inventory Optimization'.







David McPhetrige - provides a cloud-based software-as-a-service solution for optimizing the target inventory level for each inventory item in each location. David’s Right Sized Inventory SaaS solution is a data-driven, comprehensive Monte-Carlo simulation that is commercially available at www.rightsizedinventory.com.


What got you interested in inventory and supply chain?


I’m a degreed accountant. After college, my first job was in cost accounting, and I loved the relationship between financial reports and physical activities such as production, backflushes, MRP, inventory activity, etc. I quickly learned that cost accountants are the financial messengers of inventory performance – and you know what can happen when the messenger brings bad news. Also, the cost accountant is the logical interface between external auditors and operations, so I was responsible for overseeing and organizing inventory audits, justifying inventory reserves and roll-outs of capitalized operations costs in inventory and other operations-related accounting.


You describe yourself as “obsessed” with inventory optimization. What do you mean by that?


After having been through enough painful S&OP meetings, with their constant, but fruitless, arguments over inventory and service-level bad news enough times, I started thinking about why inventory seems to be such a chronic pain point. I realized that there were great tools for structuring a supply chain – tiers, nodes, etc. I was confident that perpetual-inventory systems used by businesses were mature and pretty solid. I knew that there was a reliable estimator for average inventory on hand: the simple heuristic AQOH = 50% of reorder quantity + safety stock. And it was clear to me that the challenge of this simple but dependable heuristic was in correctly quantifying safety stock, or buffer inventory.


Buffer inventory is simply extra inventory on hand to protect a target service level against random variations in demand and supply. A quick Google search will pop up plenty of spreadsheet-friendly, data-driven statistical formulas. And many ERP systems calculate a recommended safety-stock level based on forecast error. And you concluded that these time-honored algorithms were wrong?


What I couldn’t figure out was why, if these approaches work so well, inventory continues to be a pain point. So I began a deep dive into the various calculations. For instance, many formulas contain a z factor, which is supposed to convert a service level to a multiplier: higher service level = larger multiplier. This seems simple enough, but the commonly-used z factor reflects the probability of no stock-out events in a given time period, such as replenishment lead time. However, in the real world, businesses do not measure stock-out events – they measure quantities, such as units or order lines. Even the Wikipedia article on service level will tell you that an event-based calculation is harsher than a quantity-based one.. This told me that the z approach was already less than optimal, and the z multiplier would result in buffer inventory that’s larger than it needs to be.


So that’s it? The only real issue is with the z factor?


Oh, no. That’s just one of about 10 important real-world factors that are either not addressed correctly or not addressed at all in the common formulas! Here’s a non-statsy, common-sense, intuitive factor that is overlooked in most buffer formulas, and not addressed correctly in those that try to include it: Replenishment cycle, how much time goes by between reorders. Small MOQs result in frequent reorders; large MOQs cause infrequent reorders. All supply-chain people know that frequent reorders mean frequent receipts, and frequent receipts provide de facto safety stock, reducing the level of incremental formal buffer inventory. The same thing results from large MOQs, which we don’t like but can’t always avoid, with their de facto safety stock due to the large quantity on hand until perhaps just a day or so before the next replenishment is received. Though it’s easy to understand intuitively, it proved to be a big challenge to translate replenishment cycle into a correct, robust and objective mathematical and statistical analysis.






About David McPhetrige




David McPhetrige

Principal at TopDown Lean Systems

LinkedIn Profile