Supply Chain Network Optimization – How to Do it Right

The time is right to re-align your Supply Chain Strategy, for many reasons including:

• Logistics and fulfillment are the new storefronts and sales tools

• The explosion of direct-to-consumer business forces all companies to have an omnichannel strategy.

• The sourcing landscape is rapidly changing with more near-sourcing and risk minimizing.

• Sustainability awareness is increasingly unavoidable - and transportation is a big piece.

The traditional approach very often leads to more of an endless data crunch and, in best case, a mathematical answer that minimizes the theoretical logistics costs but does little to create an implementable supply chain strategy.

But you don’t do a network optimization to relive the linear programming classes from college and show your mathematical acumen. Well, some of us may. For the rest of you, here are some crucial hard-earned learning points from many network optimization projects:

1. Avoid Spending Time and Money Feeding the Monster with Data

Nobody has perfect data. A proper strategy for cleaning and curate the data will be one key to the success of the project.

The important thing is to know what data really matters and how to curate the imperfections into a usable dataset.  The most critical data for the outcome is the shipment data and it is often the hardest data to get. If you don’t have access to this data internally, the carriers do have it. They are not always keen on sharing it, but they are still the best source. The order data is usually readily available and can be used to recreate shipments though the dim/weight can complicate things. This is an area where data-enrichment from firms specializing on this can work and also item profiling to reduce the complexity to where it matters.

Most advanced models require much more data to run, but the impact of other data is less critical and can in many cases be handled with benchmarks to get a starting point and sensitivity analysis where you rerun the model with the critical datapoint varied until you find the breaking point where the recommendation changes. It is much easier to make a call when you see where it really matters and have clear choices.

2. The model will not give you a strategy. It will only tell you which alternative is mathematically the best.

Before you run the model; use the data that has been collected to profile your supply chain. This will enable you to evaluate relevant solutions.         

• Customer profiles and requirements: Delivering to the big retailers demands a different solution than direct to consumer deliveries. Two separate networks?

• Inventory profile: Certain products may have demand patterns very tilted geographically or being critical from a supply perspective. Slow-movers vs best sellers. Examples of facts that would determine the eligible alternative network structures to optimize such as Central DC, Regional DCs, Satellites, Forward Stocking Locations, etc.

• Order profile, supply profile, product profile are other examples of facts that are important to analyze pre-modelling.

3. Sensitivity Analysis

Instead of trying to create the perfect dataset. Use the model to find out with what value on critical but uncertain data that the recommendation changes. This saves a lot of time and makes the decisions relevant.

4. Use the Right Software Tool for your Challenge

The most advanced optimization tools are expensive and require a lot of effort to configure. You’ve spent a lot of time and money before you are ready to run the model. This is totally worth the investment if your supply chain is very complex and you intend to, once configured, use the model frequently. Those software tools are sophisticated and awesome with all their possible add-ons.

Most companies have a more straightforward supply chain or can optimize the network in North America, or Europe, Asia, etc. separately and then piece them together. If this is the case, the most complex tools are complicating things without the added value. Spend the time and money you save on an adequate optimizing tool to focus on strategy development instead.