Why the supply chain crisis was so impossible to predict

Online retail during the upcoming holiday season is expected to be unprecedented, with consumers spending an estimated $200 billion, according to a forecast by Adobe. But supply chains were already cracking under the pressure of elevated demand long before the airwaves were blanketed in Black Friday ads.

Bolstered by stimulus checks, consumers who tightened their belts during the early days of the pandemic have been plowing their money into buying stuff like never before. This level of demand is undoubtedly a huge contributor to the current supply-chain crisis. But experts say that even without COVID-19, supply chains would still be struggling.

A persistent lack of data about customer demand has made the supply chain feel like a guessing game — a problem technologists and supply chain incumbents alike are racing to solve.

The market for supply-chain tech is booming. According to PitchBook, supply-chain tech startups raised $15 billion in funding just halfway through this year. By comparison, in 2020 it took the entire year to raise a similar amount. This need has helped startups like fulfillment solution company ShipBob, delivery platform Bringg and freight forwarder Sennder raise hundreds of millions to digitize the supply chain.

Amar Hanspal, founder and CEO of supply-chain automation company Bright Machines, is also passionate about taking the tech-resistant manufacturing sector into the 21st century. He wants to help the industry move beyond what he calls the "industrial revolution way," where manufacturers would build "100,000 of something, put it on a container ship, have it come over and then somebody says, 'I don't need 100,000, I only need 80,000, so chuck 20,000 in the landfill.'"

Because manufacturing has traditionally lacked data, matching supply and demand has mostly been asynchronous, Hanspal said. Manufacturing doesn't have the same real-time visibility as, say, ride-hailing apps. "They know the demand. They know the supply. They know where all the taxis are. They know where all the customers are, and they can do a perfect job of matching that," he said. In the manufacturing world, manufacturers often don't know what the demand is, so it's difficult to know if they have too much or too little supply.

By providing real-time data on the production process, Bright Machines aims to provide manufacturers with the visibility and transparency they typically lack. By embedding software into the assembly line, Bright Machines can track each product as it's being built, providing information on how much supply there is at any one moment and also how long it would take to build more. Hanspal describes this as the "Domino's Pizza Tracker" of manufacturing.

But getting better data about supply is only half of the equation. Predicting demand is the much harder part.

Retailers and wholesalers in the apparel industry, for example, saw a huge whipsaw in demand pre- and post-COVID, said Jason Miller, an associate supply-chain professor at Michigan State University. Before the pandemic, there was only a "couple percent growth every year in clothing and clothing demand," Miller said. Then, during the beginning of the pandemic, demand virtually disappeared. But recently, sales have rebounded to "about 15% above where they were in 2019," he said. "You went from basically selling nothing to, now, you're selling more than you've ever sold before."

Because retailers and wholesalers had to drop their inventory at the start of the pandemic, many of them were already running lean. But the volatility of the market has made them hesitant to add excess inventory. "Retailers and wholesalers are dealing with so much uncertainty right now about what demand is going to be in several months that it makes it very difficult to want to be aggressive when it comes to ordering large quantities of inventory because they don't want to be stuck with it," Miller said.

Blue Yonder, a supply-chain management company that builds demand-forecasting software, is one of the companies trying to change this. Shri Hariharan, a vice president at Blue Yonder, thinks there needs to be a shift in how supply-chain companies are trying to balance supply and demand.

Traditionally, historical purchasing data was used to anticipate what customer demand would be like in the future. "But now, that's all been upended," he said. History is no longer representative of what today and tomorrow will look like. That, Hariharan said, means companies need to stop being overly reliant on predictions that assume there's a "very linear [and] very static supply chain."

Since the majority of business is transacted digitally, there's also more data than ever, "whether it's at the point of manufacturing or within a warehouse or at a retail store or a stadium or a hospitality location," Hariharan said. Blue Yonder uses machine learning to mine hundreds of those data points to provide its demand projections. The software combines historical purchasing data with variables like interest rates, weather patterns, seasonality and more to predict customer demand for a given time period.

Hariharan said these forecasting models are helping customers "understand that there's a new normal of demand."

The question is: Could more data and better prediction software have prevented the supply-chain crisis? It might have helped, but Miller thinks the supply chain still wouldn't have been ready for the pandemic. "People have to understand that for the most part, no system could have been built to take the shocks of the pandemic, and [...] run profitably before the pandemic," he said. Supply chains are designed to be lean and efficient, not to have extra capacity that can accommodate large shocks.

The hope though, is that as technology progresses companies can at least anticipate the next disruption — and better equip the supply chain to respond.