The Quantitative Supply Chain Manifesto

By Joannes Vermorel, last updated May 2017

At Lokad, we are discovering better ways of optimizing supply chains and we wish to help others do the same. Through our work, we have come to value that:

1. All possible futures must be considered; a probability for each possibility

Customers themselves don’t always know for sure what they will buy, when they will buy, or if they will buy at all. Uncertainty cannot be denied and should instead be embraced. Yet, uncertainty does not imply that all futures are equally probable. Some futures are more likely to take place than others. The goal of the forecasting process is to assign a probability to every single possible future. Modern computers have incredible processing power, and while assessing all those probabilities requires significant processing capabilities, this no longer represents a blocking issue.

2. All feasible decisions must be considered; possibilities vs. probabilities

Every unit of goods that you have in stock means making at least one decision per day: to keep the unit where it is or to do something else with it. Every unit that you don’t have in stock, whether because it has not yet been purchased or because it has not yet been produced, also requires making one decision per day: whether this extra unit should be “materialized” or not. All these decisions should be considered each day, for every product, for every location, for every supplier, for every route. Again, while processing power might have been an issue in the past, it is not an obstacle anymore. Hence, all possible decisions should be examined in relation to all the possible futures and their respective probabilities.

3. Economic drivers must be used to prioritize feasible decisions

Zero stocks, zero stock-outs, zero delays are only somewhat theoretical limits of your supply chain; those are not practical, feasible - and certainly not profitable - options. One key supply chain goal is to minimize the dollars of error, not the percentages of error. Thinking that improving percentages of error automatically translates into cost savings is a fallacy. Inventory costs must be balanced with stock-out costs. Purchase prices must be balanced with purchase quantities. Any optimization is fundamentally dependent on the metrics that are being optimized. In order to achieve such a business-minded optimization, economic drivers need to be introduced. Thanks to those economic drivers, it now becomes possible to prioritize all feasible decisions in relation to their expected ROI. Refining economic drivers may take as much effort as executing the optimization itself; however this is the price to pay for making sure that the results are aligned with the economics of the business.

4. Being in control requires automation for every mundane task

Automation is the key to giving management more control over their own supply chain. If taking care of the unending stream of supply chain decisions requires an unending stream of manual entries, then supply chain practitioners are the slaves of their own supply chain solution. Being required to manually supplement the solution with unending manual entries is the exact opposite of being in control.

In fact, being in control means that all strategic insights are properly factored into the millions of decisions being made in relation to your supply chain. Whenever your market situation changes, your strategic insights must be revised too. Revising a supply chain solution in order to account for the new elements in a company’s strategy should be painless, ideally done within hours, not within weeks. What’s more, there should be no limit to the amount of expert knowledge that can be injected into the automation.

5. A supply chain scientist must take ownership of the numerical results

If your supply chain is significant and has been operating for years, then preparing your supply chain data is a major undertaking in itself. Very few practitioners realize how much depth tends to be present in data, and, as a rule of thumb, a "traditional" IT department almost never does. The primary challenge lies in establishing the semantics of the data: what does data actually mean. The semantics are dependent not only on the software being operated, but also on the many operational processes being followed as well. Uncovering and documenting the data semantics requires considerable skill. Moreover, the delivery of the numerical results requires an adequate modelling of the supply chain, which in turn requires additional skills. Having a supply chain scientist take ownership of the delivery of numerical results is critical for ensuring a project's success. Without the necessary supply chain science competencies, an initiative is at risk of suffering from unidentified subtleties that may be linked to data, to supply chain processes, or to modelization artifacts. In turn, it may wreak havoc to supply chain operations once results are put in production.

---

CONTACT US