How AI Is Helping Retailers Right-size Inventory

Matt Hickman
·6 min read

No garment, no matter how durable, how locally sourced, how biodegradable, how many times it can be recycled, can ever be as sustainable as the garment that was never made.

With that eye toward sustainability, as well as cutting waste from the expense line, Rachel Liaw launched Fuse Inventory in 2016, a hyper-advanced version of the Excel spreadsheet still being used by many companies.

More from WWD

“The concept [of building supply chain software] seems very simple — you just forecast your demand and you just buy that, but there’s always disruptions, and I think the pandemic really shone a spotlight on what happens when these disruptions multiply and you can no longer just throw people at the problem,” Liaw said.

AI-driven inventory tracking is able to form forecasts from historical and current sales, price fluctuations, order quantity based on lead time, seasonality, promotions and product launches as well as market outliers such as natural disasters and pandemics. This compacted crunch of data is easily digested by SaaS, or software as a service, systems and aims to help businesses hedge how much inventory they will need and where best to sell it.

Fuse has found its greatest success with consumer packaged goods brands, particularly in beauty and cosmetics. Fashion is a far trickier inventory type to manage, though, due to its ongoing issues with size, color and style, as well as the quickly changing nature of fashion trends.

It’s a challenge that U.K.-based AI software company Sparkbox.AI is embracing head-on with its AI-based inventory model.

“We take data from retailers on a daily or weekly basis and we’re using machine learning to essentially understand how each product in each region or zone within their business will continue to perform from a demand perspective,” said Sparkbox.AI cofounder Lindsay Fisher.

“We wanted to build fashion first to prove that we could solve the harder forecasting problem first.”

Fisher said many of Sparkbox.AI’s clients to date are of what might be called the “fast-fashion” variety.

“Our goal is to help those retailers fulfill the demand in the most responsible way,” Fisher said. “If we can help people right-size their buys it means they can buy less and still grow their business.”

Other software systems are trying to corral all of the customer-facing interfaces with back-office needs — including human resources, supply chain logistics, merchandising and inventory tracking — under one software umbrella.

In recent years, this has come to be known as unified commerce — distinct from the better-known term omnichannel, which applies only to the sales portion of the site seen by the consumer. The argument for unified commerce, as Aptos’ senior director, global omni-solution principal senior director Steve Ross puts it, is the idea that the whole is greater than the sum of its parts.

“One plus one equals three-squared,” Ross said. “I’ve got clients who’ve got multiple solutions, but by having a synergy of solutions, you’re gaining more value than would a stand-alone order management system.…Unified commerce is the sum of all those systems working together and being greater than they would be alone.”

Aptos delivers AI-assisted inventory software akin to those provided by Fuse Inventory and Sparkbox.Ai, but Aptos also works primarily with larger-scale companies like New Balance, North Face and Skechers.

Dustin Jones, founder and chief executive officer of the new Unified Commerce Group, has his eyes set on smaller fish; start-ups that have been around for seven to 10 years and are valued between $20 million and $70 million. He says artificial intelligence in inventory is of “zero use” for these businesses until they grow into three or four brands, by which time he hopes to have already given them their “parachute exit” by purchasing their company.

“We do use AI for site merchandising, which is a really great tool,” Jones said. “But we’re focused on smaller-scale businesses where [inventory] demand planning is much more in the omnichannel capacity— how many small, medium, large, extra large, etc., to get right versus big company demand planning problems where they’re overbuying and it ends up in a landfill.”

JP Chauvet, CEO of Lightspeed Commerce, a Montreal-based e-commerce software company that traces its roots back to 2005 when Apple Stores first began disrupting the concept of checkout registers focuses on how inventory can be tracked and exacted right down to the point-of-sale device.

“[A client] can access order inventory levels from the POS, and what we do is automate reorder points, so we know when you’re running out of inventory and we’ll tell you, you’ve got to order 10 whatever because you’re running out of stock,” Chauvet said. “And then on the [supplier] end, they receive the order, fill the order, they ship the order to the store, and when the store receives the order, and they scan it, it automatically populates the POS with all the data [of the items contained].”

Chauvet’s company specializes in retail — mostly fashion — and restaurants. The latter, he uses for an example of how AI technology can be used to not only right-size inventory, but to right-size it for each individual customer.

“We have an insight engine and we call it the magic quadrant of menu management,” Chauvet said, adding that a chef will typically say their best menu item is also the one that sells most. “We’ll be able to tell them, ‘Oh, well if it really is your best menu item, how come every time a person returns to your restaurant, they never order it again? With data and analytics we can understand behaviors; I can see a consumer from one store to the next and I can see a returning visitor through tokenization.”

Tokenization is the encryption of sensitive data, such as anonymizing the spending behaviors associated with a credit card number, information that can be gotten through booking reservations through certain apps, or, theoretically through loyalty programs. What wine do they prefer? Are they meat eaters? Big spenders? Through their credit card footprints, diners’ tastes can be followed to other restaurants, too.

“What are the trends? What do people eat? We feed that back to the restaurants so they can readjust their menus,” Chauvet said. “If tuna belly and toro tuna is the trend of the month, we can see all the restaurants that have it are selling out of it like hot cakes, so you should put that on your menu.”

But software isn’t the only way in which businesses are using AI to right-size their inventory.

Earlier this year, American Eagle Outfitters said it was partnering with startup Radar to install RFID (radio frequency identification) sensors and tags on apparel in 500 of its stores.

With this technology, retailers are able to take inventory tracking from the Excel sheet to GPS, enabling them to know not only how much of a certain item is in stock, but exactly where it is.

“We combined the XYZ coordinate data of the RFID tags with the 3D coordinate data of where people are, so basically each sensor has these four cameras on the circumference that extract the 3D pose of people,” Radar founder Spencer Hewett said.

Hewett said he hopes someday soon to expand the technology so as to eliminate the need for checkout lines entirely.

“If you can locate every item and every person, and those people had their payment information saved on the app on their phone, and between these three signals we could charge them for what they left with,” Hewett said.

Best of WWD

Click here to read the full article.