The pandemic has provided one heck of a stress to just about every industry. One we've watched with considerable suspense: supply chains.
Would global supply chains break? No, not so far - though I'd argue most ERP-centric supply chain systems got exposed.
Back in April, diginomica took an early gut check on pandemic supply chain disruptions, via a webinar Den Howlett moderated with Noodle.ai, Navigating Turbulence in Your Supply Chain. During that webinar, I jotted down three bullet points:
- Social distancing is having an impact on warehousing and supply chain networks. Then, overlay that on other supply chain problems. Therefore: manual supply chain planning isn't viable with this complexity.
- The Noodle.ai team shared customer challenges on supply chain planning - How much is elevated consumption and how much is hoarding? How do you predict, manage, and plan for that? Data science can potentially help us anticipate these signals.
- The power of external data sets and agile supply chain planning - Legacy supply chain planning systems struggle with these data requirements. Keeping the world's supply chains running despite the adversity is the huge collective payoff here.
Since then, I've treated those as open questions to pursue. Along with an audience question that keeps coming up: can AI really help with planning amidst a pandemic it didn't predict? What are the practical use cases? That's particularly relevant given the amount of marketing snake oil around AI - a topic Noodle.ai's Gaurav Palta took up on diginomica recently (Supply chain concern #1: Unpredictability).
For another view, I sat in on a webinar from Loftware, COVID-19's impact on global supply chains. Loftware also has a dog in this particular fight, via the impact of enterprise labeling. But unlike most (unwatchable) webinars, they kept their solution talk to the proper minimum - and focused on discussions, not a slide parade. Good call, given that the CEO of Loftware, Robert O'Connor, was interviewing a very smart individual, Professor Willy Shih of Harvard Business School. Here's my top four topics from the webinar.
The pandemic exposed our weakest supply chain links, and just-in-time manufacturing took a big hit
We all know the basic timeline: China took an early pummeling from the virus, disrupting supply chains from the Chinese side. Then as the pandemic spread globally, supply chain upheaval followed, preventing the subsequent stabilization in China from offsetting logistical turmoil elsewhere.
As Shih told attendees, our weak supply chain links got exposed:
I've described it as: we suddenly lowered the level of the ocean and exposed all the rocks. We exposed all the interdependencies; we exposed our dependency on obscure suppliers somewhere deep in the tier of our supply chain.
Shih says the complexities of modern supply chains make adapting on the fly difficult. Modern technology breeds specialization, which in turn requires us to manage sub-assembly at different locations. That requires a heavy dependence on transportation and logistics - and dependable logistics are not exactly a feature of pandemic life. Example: when 95 percent of passenger flights are canceled, you've suddenly lost a huge amount of air cargo for freight hauling as well. Shih:
All of a sudden that dependable logistics that we have relied on for just-in-time manufacturing suddenly went away. All of those things exposed the complexity and the interdependencies at a global level. And for a lot of people, that was quite a surprise.
Running lean and adaptable sounds great, until it isn't. When just-in-time gives way to "where-the-heck-is-this-shipment," that's not good. Shih:
We might pull parts from China or from Europe or from Southeast Asia, integrate them into the assembly line in Mexico, and then ship it to the North American market on the assumption that I have reliable, predictable logistics.
The financial pressure has always been to have lean inventory... The immediate response is, "I might need more buffer stocks. I might need more safety stock of those critical components, especially when I only have single sources."
That's sure to be a hot topic at the next management meeting:
The question is, how does your CFO feel about that?
The pandemic also exposed product segmentation
"Personalizing" products into segmented SKUs has been hot for a while. If you're a man in your 50s, who needs vitamin C - when you can get vitamin-C-specially-formulated-for-the-50s-man? But when supply chains are compromised, SKU complexity can place an unwanted burden on forecasting. Shih:
One question that people are asking is about SKU proliferation. How much SKU proliferation do I really want? If you go look at the planograms for some of these grocery stores, you ask, "Did I really need 60 different SKUs on toilet paper?" Granted, you'll have three or four different manufacturers, including a house brand there. But do I need 60 SKUs? Because that makes it really hard to forecast... So one of the things I see coming is questioning this kind of needless variety.
That's potentially rough news for marketers who dream of new product variations to capture a bit more shelf space. It may be a welcome shift for demand planners.
The tensions between geo-political, regionalized manufacturing, and economic nationalism will complicate whatever's next
Now that the initial pandemic demand crush is behind us, you can see the splintering of agendas, from companies that want to pull sourcing out of China to regionalized distribution talk. Perhaps it's about pulling back from reliance on mega-factories, back to smaller shops. Shih raised dilemmas: are we better off avoiding dependence on large meatpacking plants with efficiencies of scale, due to the threats posed to global supply chains when they go offline? Whatever happens, it's not going to be easy to solve.
Shih makes the point that China pushed for two decades for manufacturing cost efficiencies. If you move to localized models, who picks up the higher costs? Yes, consumer behavior is changing, but few consumers are eager to take on the higher costs of "bringing manufacturing home." Regionalizing might work for some industries, but others would struggle with interdependencies. Shih cited automakers:
People are saying, "Okay, maybe I need to regionalize manufacturing, or maybe I need to go more towards having multiple suppliers. Spinning these things up is not so easy, especially when you have critical components where you have something that's very capital intensive. For example, what if you're a Detroit automaker, and I need the LCD screen for my entertainment system for my Android Auto or you know, my navigation system, okay?
Those only come from a few suppliers in East Asia. And for things like pressure sensors for my ventilators and things like that, right? So developing those alternate sources of supply is going to take some amount of time.
The other tricky part of this mix is consumer behavior. Yes, retail pandemic studies have come out, but when will the hoarding days be behind us? How do we predict the impact of colder weather, flu season, persistent unemployment, or the pace of vaccine trials, which could spur demand. When the consumer signals are this unpredictable, demand planners are not in for a fun time.
One crucial point raised by the Loftware webinar: amidst the turmoil, there are success stories. And: plenty of heroism as well, e.g. plants that quickly adapted to making essential pandemic safety products. Shih didn't mince words: those companies that responded aggressively and adapted quickly are going to come out of this in a strong position. Those that weren't ready will get overrun. Shih sees this in the education field as well: Harvard Business School went to virtual education within a week, but not all schools can say that. This is not a good time for laggards.
The next step? Document the use cases. Some companies are still shellshocked by a massive fall-off in demand, so we can't get too preachy about how next-gen tech can solve that. Worker safety will - and should - dampen output as well. How much automation can offset that is another variable. But there remains the question: can more agile approaches to supply chain planning help? If so, what do those projects look like? As Michael Hulbert, VP of Consumer Business at Noodle.ai put it, we have a chance to wrestle down the SKU forecasting problem:
One of the things we're looking at right now is the ability to see my product portfolio, and see where I have both dynamic SKU DC locations, and where it's stable ones. Unfortunately, some of the stable ones are stable because they're pegged at zero or, you know, crazy things like that these days. But, that's the first step.
What's next? A full demand signal across data sets:
The next phase... is where we would actually generate a full demand signal. Whether that's something you use for execution, or it's something you use for demand planning on more of a 12 - 18 month horizon, you're bringing to bear both internal and external data sets to run a full demand signal.
With a more agile/external demand planning process, that should mean the ability to quickly overwrite existing plans, or do exception-based changes to those plans. When it comes to predicting pandemic economy needs, that sounds a lot better than an over-circulated spreadsheet.
For us to withstand - and perhaps eventually thrive - these are the necessary conversations.