Retailers and brands have been leveraging technology such as product life cycle management tools and predictive analytics platforms for some time. But in recent years companies have increased their investments in tech focused on machine learning and data analysis. Now generative AI has stepped into the spotlight.
But how should these new technologies be used?
More from WWD
That was one of the main questions explored during a recent WWD and Sourcing Journal roundtable, titled “Getting Product Right With Technology,” which also looked at how retail and consumer behavior has changed and how retailers and brands are rethinking how they can better use data.
The session featured Danielle Schmelkin, chief information officer at J. Crew Group; Michael Appel, managing director of retail turnaround firm Getzler Henrich & Associates, and Greg Petro, chief executive officer of retail tech firm First Insight. The roundtable was facilitated by Lauren Parker, studio director at Sourcing Journal.
With AI, Schmelkin said the fashion retailer had used it for visual product recognition. She said, as an example, if a shopper is looking for a “party dress,” but J. Crew doesn’t describe any of its items as a party dress, the technology can still offer up searches of J. Crew products that match.
With the latest iteration of AI, “generative AI,” there’s been a lot of buzz in the industry that the technology could be gunning for the jobs of marketers, designers and even merchants.
Appel said merchandisers are “concerned that if AI is helping them choose a product, then maybe that’s going to put their jobs at risk. I think what you must do as management — and even in the C-suite — is to really sell the concept of utilizing AI as an enabler to help you do your job better. This is key, in terms of having companies and the associates in the companies, embrace AI as an integral part of doing business.”
Appel, who previously served as CEO and chairman of Rue21, said when he joined the retailer he and his leadership team analyzed merchant product selection. “In the analysis, we found out that 80 to 90 percent of what the merchants were picking were failures,” he said. “And it doesn’t mean that the buyers are doing a bad job; it’s just that they may not be considering all the elements that are important to the customer.”
Subsequently, Appel said Rue21 was “able to use AI very successfully in helping the merchants do two things. One is to do a better job of picking a new product. And to do a much better job of not buying products that wouldn’t have sold. In terms of gross margin performance, that’s powerful.”
When asked about humans picking product versus the technology doing it, Petro said that was a “loaded question.”
“It’s a lot to unpack,” Petro said. “The first thing to think about is when you’re doing forecasting, which is essentially what merchants are doing, they’re forecasting what is likely to happen in a probabilistic way.”
The problem is that forecasting is mired by human bias and a lack of believability in the technology. “While humans are well intended, certainly we all have a bias and consequently we bring that into our decision-making,” Petro said, noting that while AI and forecasting have been around for 20 years or so, “we’ve had enterprise forecasting models helping companies do some of that work. But the resistance points that happened during that tenure of time were really about believability. With planning tools, merchants asked, ‘OK, how do we know it’s going to work?’”
Petro said with the AI of today, it is “truly one of the big building blocks that you’re not going to be able to ignore. I think there’s got to be a receptivity in our industry to at least learn and understand AI. When a human and a computer work together, the best outcome happens.”
The discussion shifted to how the industry has evolved over time where the merchant princes have been replaced by number crunchers and statisticians who base product selection and development decisions on data. Appel said for brands to be successful today, they need CEOs who are well-versed in data science and technology as well as have the financial acumen to know how to invest in it. But they still need to have a merchant’s heart and creativity.
Schmelkin said today’s merchants need to leverage technology and data to make better-informed decisions. Schmelkin said at every company she’s worked for, she has said, “’I’m here to sell jeans in cashmere’ [for example]. My team’s not here to code. We code to sell jeans in cashmere. We understand the bottom line is: we must sell more jeans, so, what are we doing to make sure we do that? I think within J. Crew, most of our groups really think that way.”
Another change over the past 20 years has been the amount of data generated by retailers and brands. But are companies using it the right way?
Petro said there’s an enormous amount of data out there, “sifting through it and developing an actionable tactic to the information that you understand, is very, very difficult. Why? Because it can lead you in a lot of different directions.”
Petro said organizations “must have a very strong strategy and point of view at the C-suite level, about what their objective is, and then they need to work tirelessly to collect information and formulate it and structure it in a way that helps inform that decision in a completely objective manner. And what I mean by that is, even historical sales data has a tint to it because once you formulate your point of view and you execute that, you’ve sort of predisposed yourself to what you didn’t choose to do and so you miss an opportunity. And all that leads you to a point of view of, ‘Did you optimize the outcome?’”
There’s another dimension to the issue of technology, data and product development and selection, Appel said. “I also think you really have to look at it through the lens of your consumer,” he said. “And one of the things we don’t talk enough about is deeply understanding the consumer and what your consumer segments are and how they behave and what’s important to them. And it’s not just one customer. The old merchant princes say, ‘Oh, I know who the customer is.’ Well, that’s not the case. There are many customers, and the question becomes, ‘Who are the important customers? And how do they behave?’”
Appel said when retailers and brands truly understand that “and you embrace that, then everything you do is looking through those lenses. It makes the process a lot more logical, in terms of the decisions you make. I think that the proliferation of data makes it more difficult, but if you still have that overarching understanding of who your customers are, then you can get there. And that’s something that is a little bit old school, but at the same time, the research is so much more advanced today than it was before.”
Best of WWD