Heads Up… Here are Seven Supply Chain Network Modeling Pitfalls to Avoid

Heads Up… Here are Seven Supply Chain Network Modeling Pitfalls to Avoid

When considering the business rationale for undertaking a distribution network optimization, supply chain executives are first and foremost concerned with an ROI to justify the project. To achieve this and to be confident that the modeling effort will be conducted rigorously and with carefully applied best practices, executives should seek experienced network design consultants to collaborate with their project teams to interpret and consider every aspect of how industry trends will impact the development and implementation of potentially successful solutions. Even then, the most experienced network modelers can overlook these seven pitfalls that can have extremely negative effects on the process of achieving a justifiable ROI.

What to look for…

1. Freight Costs ≠ Flow Volume

A classic case that is often not handled properly. For instance, where current I/B shipments to a plant or DC have a large average shipment size on a lane but based on a scenario, those volumes decrease to the point where either ship frequency would need to decrease drastically to keep the same average size, or the rate per volume needs to increase to reflect the smaller loads.

2. Misnomer of Inventory Optimization

Multiple scenarios can be used to effectively bracket the inventory effect of various scenarios, and thereby one optimizes inventory as part of the overall cost minimization through analysis…however, inventory is almost always a post-run calculation. That is, it is not part of the objective function in the model, and is not really “optimized” as part of a single scenario. Best practice: Savvy scenario run plan development and flexible evaluations are what create conditions to optimize inventory; the models do not optimize an “inventory” variable.

3. Operational Costs Are Not All Variable

For example, labor costs (a portion, or sometimes all) are typically put on a variable cost basis, and increase/decrease in total based on throughput.  However, in the short term, even the most variable labor forces (except perhaps in a 3PL setting where clients are charged per pallet in/out, etc.) will not change headcount +/-1 x when volume changes. Understanding just how fixed or semi-fixed labor realistically is at a DC and then properly modeling those dynamics is critical to establishing accurate operational costs. Detailed XLS DC modeling, grounded in solid facility design and implementation expertise, goes far beyond what is typically represented in a network model.

4. Peak Adjustment Factors

A common shortcut for representing seasonality in a single time period model is to adjust an annual capacity figure by a formula that accounts for peaks.  While mathematically interesting and simple, the relationship between the results you get through such an approach, and what you would see with a true multi-time period model, can vary greatly.  We’ve seen significant misapplication of network capacity if the effort to create a time-phased model is not correctly structured.

5. Split Shipments

The use of bundling and similar techniques to ensure no split shipments is not a widely used practice (or used correctly). Rather, rates should reflect the characteristics of shipments (products, customers, mode, sizes) that are allowed in a scenario.  A classic flaw is to show savings from by-passing DCs and moving plant-direct for select products, not recognizing both the impact at the mixing centers (DCs) of lower shipment sizes and the true cost of the direct shipment (usually assumes same shipment profile that was achieved via the DC). Projects that have questions regarding what should go direct, and what should go through consolidation centers should be modeled distinctly differently.

6. Missed Aggregation of Product Groupings

Both product and regional aggregation can be overlooked with simplistic assumptions. Whether domestic sourced or Asia pre-built assortments for consolidation, understanding in detail how to do SKU and product groupings can be one of the most important business assumptions to clarify for the model. Creating “logistically distinct” product groups are usually not the same as “marketing” product groups.

7. Sensitivity Analysis is Not the Efficient Frontier

Endless back and forth often delays and confuses the path forward. But the executive team needs to understand more than the optimal answer from the model. Effectively guiding the project team on those factors that distinguish success, risk, complexity, and opportunity by applying “min-max-regret” and other operations research strategies – including Monte Carlo simulation – are powerful tools we can bring to help executives understand holistic business risk and opportunity.

There’s an enormous amount of data to collect and digest when undertaking a network optimization project. With this tidbit of knowledge on the seven modeling pitfalls to avoid, there should be a greater degree of confidence that the modeling process will provide the lowest landed cost and an ROI that justifies implementing the right solution.