Skip to main content

Data Science Can’t Save You: What Fashion Needs More Than Expensive Software

If the pandemic has birthed a common refrain for fashion retail, it’s that the industry is being forced to evolve seemingly overnight. Projects that required months of preparation to clear bureaucratic red tape pre-COVID now are being greenlit and cobbled together in a matter of hours, days and weeks.

And while that might be a win for getting inventory out of stores and into the hands of quarantined customers waiting for curbside pickup, CatalystAI founder Ahmed Zaidi warns that the back end of the fashion supply chain requires a bit more finesse if companies are serious about seeing real data science results and insights.

Just look at what’s going on in the education space, Zaidi said last week during Sourcing Journal’s digital event, “Fashion’s Business Model for Sustainability.” Enthusiasts have crowed about how technology would democratize education, but despite how iPads and other digital gadgets have infiltrated the classroom over the past decade, “they haven’t really cracked the problem,” the Cambridge University researcher said, because they’ve done little more than to solidify and digitize a broken process.

Don’t confuse digitization—which involves redesigning processes for digital technology—with “just being digital,” he added. Simply buying expensive new software isn’t a fix in and of itself. “Technology is just a tool” that facilitates a companywide cultural change, Zaidi continued, which for fashion should mean figuring identifying the “right ways to do innovation,” much as how Jeff Bezos alit on the twin truths that consumers will never want things slower or to part with more of their money.

Related Stories

In a time when the Googles and Apples and Facebooks of the world can throw oodles of money and perks at bright young minds with elite-school pedigrees, many have been wondering why data science grads would want to step foot inside a fashion company—especially during the current disruption. But anyone asking that question, Zaidi said, is thinking about things the wrong way. Today’s data science and machine-learning students “want to be part of a project where they can make an impact,” he pointed out. “And what they’re interested in is interesting problems.”

Where fashion companies are at fault, he added, is in failing to “articulate the problem that they’re trying to solve from a data-science perspective—mostly because they’re still trying to figure it out themselves.”

AJ Mak, founder of fashion retail-focused machine-learning startup Chain of Demand, agreed. Few if any candidates interviewing to join the tech firm’s data science team had any experience—or even interest—in retail, he countered, echoing the idea that the most “adventurous individuals” will always be lured in by the potential to solve a particularly persistent problem.

But a disconnect remains—data can give up insights but those learnings only make a difference if the supply chain is poised and prepped to react. “Supply chains are not equipped to handle data-science insights and that goes to the point that process change is just as important as technological innovation,” Zaidi said. Gathering real-time knowledge about what customers want to purchase means very little if that data can’t trickle down to the systems and suppliers that can pull the trigger on that piece of information.

Though data science can help companies predict the future, it can also be applied to optimizing decisions, Zaidi noted, or reducing the uncertainty around a prediction. Instead of ordering a total number of garments from a supplier, an apparel firm might hedge its bets by transferring some of the remaining commitment risk into raw materials and capacity that could inform another product should demand morph, he explained. “That’s the happy medium between having complete on-demand manufacturing and the current long lead times that we have,” he said, enabling flexibility to react to fluctuations in demand.

Chain of Demand, meanwhile, quick rolled out a pre-season modeling tool once it realized that 90 percent of its fashion clients were ill-equipped to act on the insights borne from its in-season model designed to “help them replenish and make in-season moves,” Mak said.

“It’s very hard for us to prove our value” if a client can’t translate Chain of Demand’s insights into action, he added.

The company helps apparel and footwear firms tackle the twin thorns in fashion’s side: unnecessary markdowns and excess inventory. Mak said Chain of Demand has helped clients reduce their markdowns by up to 15 percent while some have trimmed their year-end inventory by 50 percent, improving cash flow and the cost of capital. “That’s what gives us confidence in saying we can help improve profitability and sustainability at the time same time,” he added.

Making smarter bets on the supply side has similarly beneficial benefits, according to Zaidi. Optimizing when and how to commit resources to production and raw materials can improve working capital cycles, he added, “which can have immense benefits for the company on the margin side but also reduce the amount of inventory that you produce, and obviously has huge impacts on the environment and sustainability.”