“Baseline Validation” is as elemental to a network modeling workplan as “creating good jobs” is to a candidate’s economic plan. It has to be there. You can’t be against it. But it’s not always clear whether there is any real meat to that part of the plan.
A baseline scenario is the first model run after completing the build-out of the model. You use a recent historical demand and force the production flows through the plants and warehouses to match the historical actual from that time period. In the validation of the baseline, you check the output to ensure that the modeled results are highly comparable to the actual. In this way you build confidence that other scenario solutions can be trusted.
Who can be against that? Not me. I only bring this subject up because I believe what clients expect and receive at this stage is not enough. Too often, a mid-project meeting is held where cool maps depicting the baseline flows and summary costs are presented. Everyone nods solemnly and the consultants proceed to the fun part of the project that include optimization, scenarios, and sensitivities.
The dirty secret is you can’t be confident of a model’s legitimacy solely because the baseline costs make sense. Other factors may compromise the validity and must be guarded against. The following are a few precautionary steps that might improve your confidence in a model before testing scenarios or running wide-open optimization:
Validate the Constraints: .
Many asset driven models have critical constraints that might not be violated in the forced baseline but still are not accurate enough for testing alternative scenarios. If important, are production and warehouse capacity utilizations close to the actuals? Have key limits on complexity been verified (e.g., # of products produced by a plant or line)?
Validate the Cost Logic:
It’s easy to get valid baseline costs for historical flows. You know the volumes and you know the costs, so creating cost lanes for these movements could be very simple. But creating a set of cost rules that apply to both historical and potential and actual lanes without bias toward either is a well-known challenge. A proper model validation scrutinizes the rules and assumptions that went into building the production and transportation cost rules. These rules should be approved by management before any results/suggestions beyond the baseline are shared.
Validate the Non-baseline Flows:
To say the feasible flow of products through alternative (non-baseline) sites needs to be assured is not exactly an earth-shaking proclamation. But this isn’t always a no-brainer, particularly when you are testing suppliers or plants in new oversees regions. With possibly new modes of transport (e.g., ocean freight, rail) and new flow rules (movement through ports), and new constraints, the potential for misstep is enough to take extra precautions and carefully verify the viability of these flows in your model set-up.
The list goes on and depending on the type of modeling engagement, the focus will vary. This is not a commercial for a 26-week project plan. There is high likelihood that the modelers are already testing these concerns and simply not boring the executive committed with all the detail. Probably.
If you want to end up with a valid model, not just a valid baseline cost, follow the advice from President Reagan and “Trust, but verify”.
(This is a guest post from Tom Nicholas. It first appeared in Supply Chain Digest. Tom has his own independent consulting firm but also works part-time with Opex Analytics. He has been doing supply chain and manufacturing modeling for many years and is an expert in the CPG industry. Prior to starting out on his own, Tom was a Senior Manager with PwC’s Supply Chain Strategy Practice. Before joining PwC (formerly C&L) in 1996, Tom spent six years with PepsiCo, working at Frito-Lay in production management and at PepsiCo Food Services (PFS) managing Logistics projects. His recent consulting engagements include supply chain strategy development, network optimization (plant rationalization), sales and operations planning (S&OP), inventory analysis, and make-buy analysis. He has earned his MS in Supply Chain Management.)