While the 20th century contended with - and effectively conquered - production automation, the 21st century is grappling with different classes of complexity entirely. Unlike the previous century, whose entropy1 was largely confined to physical constraints, modern supply chains exist in a far greater state of flux. This flux encompasses the same physical challenges of the past 100 years (e.g., responding to natural disasters), but is further compounded by the stochastic trends and consumer demands produced by increased globalization and technological advancements. Accurately diagnosing the scope of the challenge ahead is the first step in effective supply chain optimization.
Watch the Lecture
Better User Experience (UX)
Consumer habits and expectations tend to evolve with advancements in available technology. A good example is the increasingly sophisticated telecommunications of the 1970s, which in turn facilitated the emergence of call centers and, by extension, ushered in the era of telesales. From a supply chain perspective, this was extraordinary, and represented a somewhat beta version of the complexity we see today2.
50 years on, e-commerce has taken this complexity and increased it by untold orders. Though products are often still delivered by hand, the method through which orders are placed, processed, and tracked is exponentially more intricate. Same-day delivery, to take just one example, increases the overall complexity of satisfying customer orders by amplifying the supply chain considerations.
Upstream, one must allocate sufficient resources to this endeavor, namely delivery personnel, vehicles, and equipment. Downstream, the shortened time horizon for delivery means expedited processing, picking, packing, and shipping of orders, which necessitates routing optimization as well as training and/or equipping staff with requisite GPS-technology. This is to say nothing of the additional forecasting entropy introduced by same-day delivery.
In the previous century, industrial patterns predominantly centered around mass production, and this limited variety made optimizing production and supply chain processes a far less complicated (though certainly not easy) task. Fast forward, and modern configurability options allow consumers to carefully calibrate their purchases to a degree practically unimaginable a century ago.
While this is certainly a boon for customers, it augments the overall supply chain entropy in several ways. Beyond the increased demand forecasting difficulty for the individual SKUs in a configurator3, quality control and order fulfillment become exponentially more complicated as consumer-optionality increases.
An additional class of supply chain complexity is the series of programmatic options one can leverage to navigate the entropy described above. Though these are designed to assist the supply chain practitioner, each option introduces its own suite of considerations. Some examples include:
Cloud 3PLs: Cloud-based third-party-logistics and storage, such as Amazon’s FBA (Fulfillment by Amazon), can provide increased flexibility and reduced infrastructure costs for businesses.
However, these services are primarily designed to be operated using APIs sitting atop the client’s preexisting enterprise systems, which can present integration, compatibility, and adoption issues4.
Autonomous vehicles: Though this is still a relatively nascent technology, the long-term viability of autonomous vehicles in supply chain is evident. Automated guided vehicles (AGVs) and autonomous mobile robots (AMRs) not only reduce human error in transportation but can also be deployed to automate certain warehousing functions, such as picking and packing.
Much like 3PLs, there are a raft of infrastructure and adoption speedbumps to clear, though the future is very bright on this front5.
Predictive maintenance: Given electronics have become increasingly more affordable, modern machinery can be outfitted with thousands of sensors whose purpose is collecting data on the performance and integrity of the machine itself. This data can be scrutinized - again, through automation - to proactively identify issues before an adverse event strikes.
The aerospace sector is a notable example, where it is commonplace to install sensors on planes. These sensors track data across thousands of flight hours, which is analyzed using machine learning algorithms to detect signals of potential failure. An Airbus A350 has as many as 50,000 such sensors, and the data collected not only reduces costs and downtimes but potentially saves lives6.
Supply Chain (d)Evolutions
A regrettable vector of chaos is the world’s occasional tendency towards disorder in and of itself. Unlike the entropy described previously, where increased supply chain disorder was an unfortunate result of positive evolution, this brand of entropy is the low-water mark of human invention, i.e., supply chain complexity through devolution.
In plain terms, these are instances where supply chain complexity increases without any tangible benefit, and usually through errant intervention. These harbingers of mayhem include, among others:
Social networks: Despite presenting myriad marketing opportunities, online networking platforms can introduce additional and accidental volatility, such as products becoming overnight fads and experiencing worldwide demand7.
Conversely, a client’s reputation (or the reputation of a key supplier/consumer) can be utterly ransacked by a social media dogpile in the space of a few minutes. Either of these digital events (to name only a few) can wreak havoc on a company’s supply chain.
Government regulation: In the mid-20th century, US companies were subject to approximately 2,600 pages of regulations; nowadays, that figure has swelled to over 200,0008. Given the geographically dispersed and interlocking nature of supply chain, federal action in one jurisdiction tends to ripple throughout the system.
These intercessions can be as unforeseen as they are swift and devastating. For example, a factory in Shenzhen closing due to a local lockdown can cause supply chains in Seville to slip into complete disarray.
Bloatware: Software is a blessing for logistics and is the very DNA of quantitative-supply-chain, but that does not mean that all rising ships are seaworthy. Vendors tend to continually add features and capabilities to their products in order to sell new versions and upgrades.
This leads to software becoming increasingly complex, sometimes to the point of collapsing under its own weight9.
Philosophically speaking, the complexity one encounters in supply chain can be viewed through two distinct lenses: that which is accidental and that which is intentional. The former is often man-made and can be reduced by having the courage to cut through unnecessary bureaucracy and inefficiencies; the latter, however, tends to be an inherent characteristic of a system, typically requiring superior technology.
Accidental complexity includes the slow and steady accretion of redundant communication channels in day-to-day business, such as tedious, virtue signaling emails and meetings. These may seem trivial, but the opportunity cost of wasted resources and bandwidth gradually adds up10.
This class of complexity is a bug and not a feature, hence it can, by and large, be eliminated through judicious management alone.
Intentional complexity, in supply chain terms, encompasses all the factors that are inherently complicated. The very foundations of supply chain, for example, are mastering the optionality, variability and flow of physical goods across the vast and distributed supply chain network.
These complexities persist regardless of how uncluttered one’s Google calendar is. They are, by definition, complex qua complex and, unlike instances of accidental complexity, are hurdles that cannot be cleared through will alone. They must be tackled with suitable and superior technology.
Entropy is a measure of the degree of disorder or randomness within a system. High entropy indicates a system is disordered; low entropy indicates an ordered one. Imagine a deck of cards that is neatly stacked in ascending value, with the suits arranged in alphabetical order. One could say this deck has a relatively low entropy rating (expressed in joules per Kelvin, were this a real example). That same deck, now shuffled, would have a significantly higher entropy rating, given the increased randomness. If one were to toss the deck into a stiff breeze the entropy would, as you can imagine, jump even more. ↩︎
Telesales introduced large-scale remote ordering, which complicated supply chain management by increasing the need for even more accurate demand forecasting, efficient inventory control, and timely order fulfillment. The shift from in-person sales to phone-based transactions also necessitated robust logistics infrastructure and reliable delivery services to ensure customer satisfaction and maintain a competitive edge in the market. ↩︎
Also known as choice boards or design systems, these online mechanisms assist consumers in the configuration process, such as when customizing a computer order. ↩︎
An API - application programming interface - is a set of rules and protocols enabling software components to interact. It is the bridge between one’s enterprise software, such as an ERP, and the 3PL’s own interface (the API), used to facilitate the exchange of data and information. ↩︎
Safety and security issues, government regulation, and social acceptance are three immediate challenges to widespread adoption. ↩︎
The integration, data management, and overall skill curve for predictive maintenance is steeper than the previous two examples in the section. That said, depending on one’s sector, the potential long-term value is rather difficult to overstate. ↩︎
Going viral, in the parlance of our times. ↩︎
Figure taken from ‘Regulations’ section of plainlanguage.gov. Note, this data does not include state and local regulations, or the guidelines of additional regulatory agencies. This is purely the general, enduring, and overarching framework set out by the federal government. This is not presented as an inherent negative, rather an indicator that the trend is manifestly tipped in the direction of increased oversight (for better or worse). ↩︎
As supply chains rely on multiple interconnected software applications (APIs, ERPs, etc.), this, in turn, can create a flabby bloatscape. ↩︎
Wasted time comes in many flavors, but the net result will be the same. A mathematical mindset is helpful here. A single pointless meeting per day (lasting, say, 20 minutes) amounts to almost 80 wasted work hours per year or two full work weeks (assuming a typical US work year with a charitable 4 weeks of vacation). ↩︎