Probabilistic MEIO in tandem with Multi-echelon Replenishment Planning deliver Service Levels.

The primary objective of multi-echelon replenishment planning is to synchronize inventory levels and orders across various echelons and stages, aiming to optimize total inventory costs while upholding high service levels. This necessitates optimizing replenishment decisions while considering multiple trade-offs. In single echelon planning, as seen in the previous edition, planners often struggle to balance interconnected trade-offs and commonly resort to building excessive buffers to tackle Bullwhip effect. They rely on inefficient methodologies (refer to the episode 'What's wrong with ABC Inventory Classification') and tools for decision making. However, in multi-echelon planning, the solution handles bulk of these parallel trade-offs, allowing planners to focus on managing boundary conditions and making value-added decisions.

From an end-result standpoint, multi-echelon inventory optimization and multi-echelon replenishment planning appear to serve the same purpose. However, the difference lies in the way results are achieved.

Some solutions integrate inventory planning and replenishment planning into one planning run, treating inventory planning as a consequence of replenishment planning. These solutions typically use objective functions aimed at minimizing costs or maximizing profits. Consequently, they may prioritize certain types of orders over others or fulfill orders that drive sales at higher costs. Such solutions often rely on linear programming (LP) engines, which normally struggle to accommodate more than a few variables and constraints. As the number of constraints and boundary conditions increases, the running time of these engines can grow exponentially, and sometimes resulting in no feasible output. For instance, incorporating additional product segments or market segment-level service optimization, or specifying a minimum service level for a particular segment type, introduces additional constraints that can significantly increase planning run times.

Moreover, these solutions are unable to recommend safety stock norms and service levels for individual SKU-locations before running replenishment planning. As said before, this is because inventory norms are derived as a consequence of replenishment planning, and there is no pre-replenishment inventory or service level recommendations available for review to facilitate corrective tactical adjustments of working capital. Such solutions with linear programming (LP) engines often remain as black boxes, making it challenging to explain the results they produce.

In contrast to LP-based solutions, probabilistic solutions are considered more advanced. In these solutions, Multi-Echelon Inventory Optimization (MEIO) and network balancing norms are set and are adopted in the anticipated replenishment decisions. The outputs typically include SKU-location level stock recommendations as well as market, product-segment level inventory and service level recommendations.

In probabilistic MEIO, inventory optimization occurs before the replenishment run and encompasses Mix, Stage, Lot Size, and Postponement optimization, all simultaneously within a probabilistic framework (refer to a previous episode on this). By integrating the opportunities from all optimization types with the accuracy under uncertainty provided by the probabilistic approach, these solutions can achieve maximum efficiency. Another advantage lies in flexibility and planning runtimes, where multiple scenarios involving inventory versus service level decisions can be derived quickly.

While an LP based, or a non-probabilistic mix optimization, may typically achieve around 30% of the potential gain over traditional safety stocks calculations using conventional formulae, a standalone probabilistic mix optimization alone can achieve around 90% better results, while the combination of all optimizations done simultaneously can extract the full 100% potential gain (Reference: Stefan de Kok’s article titled- “Safety Stock vs Inventory Optimization).”

Probabilistic MEIO which offers true inventory optimization, enable businesses to facilitate informed decision-making and foster collaboration among Sales, Supply Chain, and Finance teams. By leveraging shared information, these teams can establish Inventory Policies and seamlessly implement agreed-upon working capital strategies into the model, translating high-level information into specific Service Levels and Safety Stocks for every Item at each Location. All decision making made possible before running replenishment planning.

It's important to emphasize that common Sales & Operations Planning solutions cannot provide maximum optimization. They are process enablers and function at the aggregate levels to balance demand-supply tradeoffs, but sacrifice granularity in inventory optimization and service level improvement opportunities.

Next week, let's center this discussion back to multi-echelon replenishment planning and bring this discussion to a conclusion. Meanwhile please feel free to leave your opinion here.

To be concluded...