Everything I’ve Learned About Supply Chains

Fragmentation is a feature, misaligned incentives cause poor data, and supply chains are just fine

Andrew Nguonly
9 min readJan 25, 2024
ChatGPT 4 prompt: “Create an image of a fragmented supply chain. The supply chain is dominated by “big players” and the “small players” have little influence. There’s a lot of missing data. Even though there are a lot of external disruptions to the supply chain, things seem to be moving along just fine. Everything is slow, but ok. Generate the image in the style of Salvador Dali.”

Mismatched Expectations 🧐

Three months ago, my co-founder and I decided to wind down our sales co-pilot product and pivot into learning about supply chains. The domain felt new, interesting, and ripe with opportunities given all the headlines in the last few years about trade disruptions and empty grocery store shelves. We thought we could bring a fresh lens to the space and apply some of our experience optimizing digital pipelines (e.g. data pipelines) to the physical world.

After some initial research, we formed a few ideas about how supply chains could be improved. In retrospect, our ideas were obvious and maybe we were naive to think that we could come up with something original after just a couple of weeks of Googling. Still, the issues we discussed felt unaddressed by the market. Over the last sixty days, we interviewed dozens of supply chain professionals across a handful of industries and functions to get the lay of the land. We spoke with truck drivers and Chief Supply Chain Officers. Sourcing specialists and network optimizers. Engineering architects and supply chain analysts, all at varying levels of experience and seniority.

What I learned did not match my expectations. Before I proceed, I should start by saying that these are my views as an outsider. I have no skin in the game. I have no experience working in supply chain. My comments and ideas are prefaced with the utmost respect and humility to those who are actively working to keep our grocery store shelves stocked and deliver our packages on time. It’s impossible to know what’s going on unless you’re working day in and day out to keep the world running smoothly.

Here’s what I’ve learned…

Fragmentation is a Feature 🧩

Follow-up ChatGPT 4 prompt: “Make it more fragmented. Make it have PDFs, emails, CSVs, SMS, EDI, API, and phone calls, all interconnected.”

It didn’t take long to realize how fragmented the software and data ecosystems are for supply chains. Many companies adopt multiple software solutions to meet precise requirements, which inherently leads to data synchronization challenges. Procurement software syncs with the ERP. The ERP pulls updates from 3PL integrations. Multiple warehouse and transportation management systems (WMS and TMS) are accumulated through mergers and acquisitions.

Data is produced in many formats: phone calls, SMS, email, CSV, PDF, API, EDI, and the list goes on. It’s synced bidirectionally between Point A and Point B and unidirectionally from Point C to Point D. Some data is synced every day, while other data is synced once per week. Some data (e.g. cargo tracking) is simply unavailable. The level of fragmentation is like no other.

A Unified Orchestration Platform 🏆

This brought us to the perennial problem of accurately counting inventory. Ideally, the count of inventory in a software system should not only match the physical count in a warehouse, but should also match the count in derivative systems that sync with the source of truth. Naturally, we thought that moving towards orchestration would be a helpful approach to closing the gap on accurate inventory numbers. We proposed a “unified orchestration platform”, a software architecture that would redefine how data is moved throughout disjoint supply chain systems. Instead of a bespoke model for data synchronization, data should be proxied through a top-level orchestrator and only the orchestrator could decide when and where to push the data next.

We quickly received feedback that the effort to migrate to the new architecture would simply be too difficult technically and may not have the desired ROI. Selling the concept to leaders and decision makers was equally hard. This was a reality check that we conceded to. We learned from a supply chain architect at a Fortune 500 retailer that some companies don’t even bother modeling exact inventory counts, but instead compute perpetual counts. Alternatively, some companies resort to manual counting or build in enough buffers to withstand errors to a certain degree.

The same individual also shared that in some cases, “unified orchestration” is not desired at all. For example, it may be desirable for warehouses to be able to operate independently from the broader network. In this scenario, synchronizing inventory counts from a warehouse to an ERP might be delayed, thus breaking the concept of a top-level orchestrator.

Laws of Supply Chain Physics ⚖️

After confirming the doubts with other supply chain engineering leaders, we let go of the idea. In Sam Altman’s post titled “What I Wish Someone Had Told Me”, he urges readers, “Don’t fight the business equivalent of the laws of physics.” Fragmentation feels like one of those laws for supply chains. It is a feature, not a bug. Don’t fight it.

Don’t fight the business equivalent of the laws of physics — Sam Altman

I expect that the number of software solutions will increase as the needs for supply chains become more specialized and niche. Because of this, a new solution should embrace fragmentation. It should operate with the lowest common denominator of data formats (e.g. phone calls, email) in addition to modern formats (e.g. JSON). Large language models (LLMs) may be a natural solution to smoothing out differences in data format. It seems obvious that LLMs can act as the glue between all of the systems in play. However, I caution that the complexity of a viable solution must not scale with the fragmentation. In other words, if fragmentation is increasing, complexity must remain flat.

Palantir Foundry was the closest solution we found to a “unified orchestration platform”. It’s hard to see how a mom-and-pop manufacturer could adopt the Ferrari of data platforms. Conversely, how can we expect a large retailer with limited engineering support and legacy software systems to migrate to an unproven paradigm without interrupting the flow of data? It’s a tall task and maybe one not worth it.

Misaligned Incentives Cause Poor Data 🏚️

Follow-up Chat GPT prompt: “Make the data quality look really poor. Sometimes, there are “black holes” of data.”

By now, we all understand that incentives drive behavior. Understanding the incentives that drive supply chains was our next goal. We wanted to understand if there was an opportunity to build an end-to-end supplier transparency network, a data network that would unlock all information between nodes in a supply chain. Is it possible to see the metrics of your supplier’s suppliers? What drives a factory to accurately report schedules and inventory? When the world is at the behest of China for top-tier manufacturing at the cheapest price and fastest turnaround times, what incentive do they have to share information?

Early Data, Late Pick Up 🚚

Almost everyone we spoke to mentioned data availability and data accuracy as one of the biggest obstacles for tracking the movement of goods. Companies like FourKites have made tremendous progress in unlocking visibility into supply chains, but there are still significant gaps. Despite the efforts with blockchain and the decreasing costs of IoT tracking devices, lack of data is still a limiting factor for jumpstarting efficiencies.

The process of collecting data presumes that the act of producing data is even possible. And producing data presumes that there’s sufficient motivation to do so. A truck driver who arrives at a port may update their status as “arrived”, but it may be the case that there’s an hours-long line to unload cargo and pick up the next load. The driver is under pressure to report “on-time” pickup, so what incentive do they have to report a delay? After all, they are technically at the port. To be clear, I’m not blaming or pointing fingers at truck drivers. They are the heart and soul of our supply chains! I’m simply pointing out a scenario where an individual node, acting in their own best interest, may produce a suboptimal outcome for the rest of the nodes in the chain. Incentives are not aligned.

Large Players Get Their Way 🐘

We spoke to a few leaders in strategic sourcing roles at small startup companies and one of the dynamics we discovered is that the business of sourcing is actually the business of selling. Small players who rely on manufacturing or supplies from large players often don’t get their way. This might manifest in the form of less preferable contract negotiation terms or a lack of updates and data. Large players have all the leverage in the supply chain. There’s simply little incentive to “work down”.

As a result, sourcing teams must sell their company’s vision to the larger player. Once a deal is made, they ask for updates, but emails go unanswered on the first try. Sometimes, updates are never received until items arrive at the warehouse. Finally, when data is provided, it’s not in a format that’s compatible with existing systems. Onboarding suppliers can be one of the most time consuming tasks for a sourcing team. A robust supplier integration platform is needed, but there will always be a long tail of suppliers who simply do not integrate. Data must be handled manually.

Breaking the Chain ⛓️‍💥

The challenge of aligning incentives for every node in the supply chain is exacerbated by the fragmentation feature. Assuming that fragmentation is increasing and every node in the chain continues to act in its best interest, how do you break this cycle? This may be the biggest of all challenges in supply chain.

Our original idea for an end-to-end supplier transparency network was dependent on a recursive approach to addressing data needs that would trigger an unlock and break this cycle of data obfuscation. Instead, what I’m realizing is that the vast size and scale of the global supply chain trounces any individual incentive to provide for the greater good. There are too many players and too many needs.

Supply Chains Are Just Fine… 😕

Are they? Every other day, I come across a new article or Twitter thread from Ryan Petersen stating how badly supply chains have been disrupted. I believe what I see and what I read, but something isn’t quite adding up.

Throughout our interviews, I observed a thread of calm, cool, and collectedness — a milder temperament than the individuals we spoke to when building our sales co-pilot. Apologies if I’m stereotyping. This is just my interpretation from an extremely small sample size. Calm and collected is a good thing.

Right?

In some ways, I wish I felt a bigger sense of urgency from the interviews. Maybe even a panic. In the voice, in the tone, in the disdain for the status quo. When interviewing sales development reps and account executives, we’d often hear how much they despised using Salesforce. But not once did I hear anyone express such contempt for the current state of affairs of supply chain software. It seemed like supply chains were just… fine.

Bad Times 😖

The supply chain industry is down badly right now. We spoke to an owner/operator truck driver at the Port of Oakland one Friday afternoon and he mentioned he hadn’t been paid by his broker for nearly two weeks. He shared that with us as if nothing was wrong, as if there were no consequences and life goes on. He was calm and jolly. But in my mind, I thought, how can this be okay?!!!

A startup founder building in supply chain told us that during bad times everyone looks bad, but no one gets fired. Companies ride out the storm and do their best. This point moved me. I started to press others about the comment. What I learned is that there’s shrinking accountability in a system with a sprawling number of parts. There’s no one to blame, no one to fire. I’m not in favor of firing people for trivial mistakes or minor accountability details. My point is that there seems to be some level of contentment or helplessness that may be hindering innovation and progress in supply chains.

New Energy 🚀

I can’t deny that supply chains have been significantly impacted over the last few years. The Red Sea is under attack and shipping prices have increased significantly. Convoy’s business seems to have disappeared into thin air. Supply chains are not fine!

We spoke with a Chief Supply Chain Officer at a small consumer brand. He shared that he came from a finance background and had moved into the role very recently. The conversation was refreshing. I could tell right away that he was bringing a beginner’s mindset to the role — open to ideas and eager to learn. This energy is what’s motivating me to continue learning. Innovation in supply chains will take time. But time is not enough. Change will require a new generation of outside thinkers working hand-in-hand with seasoned professionals to push our supply chains to new levels of resiliency and efficiency. One day, maybe we can break the chain of misaligned incentives.

If you’re a supply chain professional and interested in chatting, please reach out. If you disagree with some of my comments or views, definitely reach out! I’d love to chat.

Follow-up Chat GPT prompt: “Make the supply chain perfect. Technology, people, and the physical world are in perfect harmony.”

--

--