Yesterday, I said that Supply (Chain) Network Optimization (SCNO, or SNO, but not snow), which is not easy, is the optimization of your global distribution network to minimize costs while controlling risk to an acceptable level. Furthermore, it’s not something you should tackle as part of your everyday sourcing project.

[

Yes, I’m saying it again. And the next person who dares to suggest otherwise gets an e-coupon for a free smack upside the head, redeemable the next time we meet. The nature of distribution network optimization is that it cannot be optimized within a single sourcing scenario, and any attempt to do so is likely to do more harm than good. To truly optimize your network, you have to optimize across all of your buys, and even in any given year, you’re likely renegotiating less than a third of your major contracts and a quarter of your buys, and you don’t expand into new countries overnight.

Furthermore, since you are contracting for fixed rates, not fixed costs, and your total costs depend on actual volumes, which are forecasted and variable, your freight costs at the time of award are approximations. If you’ve done your homework, and are revisiting your freight contracts semi-annually or annually, like you should be, you should be within a few percentage points, at most, with these estimates. Let’s say worst case scenario where you’re off by 5% because you didn’t quite meet demand to get that rebate you were hoping for (since your estimate demand was only 1% above the cutoff). In an average strategic sourcing scenario, you’re looking at freight in the 5% to 15% range, so you’re off by at most 5% of 15% or 0.75%. Worst case, you’re off by 0.75% – but since you have an equal chance of being off either way if you’re estimating properly, it’s going to average out over all your awards and you’ll be within a fraction of a percent of optimal (unless oil spikes again and huge surcharges come into play, but since all your competitors will be paying the same surcharge that every carrier will be levying above the board, you’ll still be ahead compared to your peers, relatively speaking, if you have the right lanes at the right volumes for the right rates with the right carriers).

]

The proper way to optimize distribution network costs is to semi-annually or annually analyze all of your projected transportation needs over the next 6 to 12 months using all of your projected shipments (based upon current contracts, forecasts, and current patterns), aggregate volumes across lane groups (defined as the set of lanes that take a product from region A (such as a set of posts on the southwest coast) to region B (your major re-distribution center outside Chicago), bid out the appropriate lanes or lane groups to one or more carriers, and optimize a transportation award to these carriers who quote rates based upon minimum volume guarantees (such as 75% of expected volume across the lane). This gives you highly accurate freight rates to use in your award allocations.

On top of this, every one to three years you should be re-optimizing the flexible aspects of your distribution network. By this I mean, re-evaluating warehousing space that you are leasing or that is highly liquid and could be easily sold, re-evaluating the air and ocean freight options to you, re-evaluating the ports you are using, and re-evaluating your shipment consolidation strategy (should you always wait for shipments from multiple suppliers to fill the container or should you use a third party that can consolidate shipments for multiple buyers to fill the container).

For example, maybe sales on the west-coast are high and the east-coast are low, maybe you’ve experienced three major delays at your current port of choice, and maybe part of your cargo is always depreciating by waiting for enough cargo to fill a container. In this case, you would model the expected costs of retaining your current distribution network versus the costs of different network designs that cancelled the lease of one of your east coast distribution facilities, used a different port, and / or used a third party logistics provider to determine whether or not a redesigned network could offer you a lower operating cost (taking into account the one-time fixed costs associated with any restructuring you had to do, amortized over appropriate minimum contract periods).

Then, every three to five years, you should be re-optimizing the distribution network as a whole, including the fixed aspects of your distribution network. By this I mean every single warehouse and facility, every single warehouse and distribution facility, every single port of exit and entry, and every single carrier is considered to be floating and the cost of the current network is compared against a fully optimized solution (that takes into account all of the one-time fixed costs that will be associated with any and all restructuring) optimized over different, appropriate, minimum contract periods (from three to seven years, as the optimal model may change from year to year as volume projections will change year over year and existing contracts will expire, decreasing the fixed costs associated with a restructuring). Then, each of the optimal network model solutions that represent a significant cost savings should be compared against the current network model and those that represent the minimal deviations from the current model analyzed in depth. (Even if a totally redesigned network could offer lower cost savings, it’s a bad idea to re-design your whole network at once. Most big-bang efforts end in failure. Furthermore, it usually only takes a few major changes to produce significant savings, if done, right, and the longer it takes to redesign your network, the more you are relying on the accuracy of your long term forecasts to realize these savings, increasing the risks associated with redesign.

Finally, at least the first few times you do this, be sure to bring in an expert and make sure the appropriate tools are available. Even though these resources may be expensive, partly due to the relative lack of expertise in this area and partly due to the significant education and experience required to do these projects properly, (since you need expertise in Operations Research (OR), optimization, modeling, supply chain, analysis, and consulting), as another resident guru of the blog-sphere would be quick to point out, it’s not what you pay, but what you get out of it. You might think that 3K to 5K a day is expensive, but if that person, with access to the right tools, saves you millions of dollars, a hundred grand or two in professional services, (on-demand) software costs, and project costs is paltry in comparison. It’s the net value of the ROI, not the cost, that’s important, as it is in any business project.