The risks and benefits of AI take centerstage at Shoptalk
When a team of ThredUp engineers debuted a new AI-driven search tool at an all-hands meeting in January, the company broke into applause. For the past seven years the company had been building its search function, but the new AI-driven tool was stood up in seven weeks, recalled CEO James Reinhart.
“I’m probably one of the few people you talk to who thinks AI is underhyped,” Reinhart said.
ThredUp’s Reinhart was one of hundreds of CEOs who attended Shoptalk in Las Vegas this week extolling the benefits that generative AI can have on the e-commerce industry. Companies like Walmart, Lowe’s and Domnio’s shared how they use AI to improve in-store processes for associates and customers, while Google’s head of commerce Maria Renz discussed how AI can be used to personalize shoppers’ experiences.
But none of these advancements come easy or cheap. Small- to mid-sized brands have to wrestle with whether it’s worth spending money on new technology, while larger companies like ThredUp have to pull together dedicated teams to develop their own systems.
But the buzz — and the budgets — are there. A Salesforce survey released with the Retail AI Council during Shoptalk showed 81% of respondents have a dedicated AI budget, with about half assigned to generative AI. It also found significant roadblocks to advancement due to the issues around collecting clean data to feed AI models, with nearly half struggling to make their data accessible.
Some Shoptalk attendees urged caution around AI tools. Colleen Aubrey, svp of advertising and tech at Amazon, set a baseline for the conversation in an opening keynote on Sunday afternoon. While there’s “enormous amount opportunity ahead of us,” Aubrey said, there’s “significant hard work” to be done to ensure consumer-facing AI is accurate in terms of product description and representation. “There is no room for misrepresentation,” she said.
Those who get into adding AI tools into their business should be looking for the places where “it can add distinct value,” Aubrey said. In its e-commerce offerings, Amazon has deployed generative AI tools for sellers to improve their product listings, as well as an image generator rolled out in October.
“Keep in mind it’s not the only technology,” she said.
But at a tech-focused event like Shoptalk — where vendors hawk software and brands look for leads — many heads were drawn to tools that could be plugged into a company’s operations. Modern Retail spoke with Attentive CEO Amit Jhawar ahead of its announcement of new AI-driven SMS tools, AI Journey and AI Pro, which he showcased during a Tuesday keynote.
The text marketing service has piloted generative AI to send marketing texts based on two trillion data points from the past 10 years. In addition to personalized messages, there’s also an AI-driven audience-building tool. Brands like Reebok, Brooks Brothers and Forever21 are among the 100-plus beta testers that have used the services. So far, top results include seeing 120% more clicks and 115% more revenue in purchases compared to human-crafted campaigns, Jhawar said.
But getting the tools rolled out wasn’t an easy lift. It took at least four tries in the past year to come up with the current model that leads to higher conversions than human-crafted messages, Jhawar said.
“It’s a whole new capability,” he said. “You have to have a very dedicated effort and we’re investing multiple tens of millions of dollars behind this effort.”
Jhawar also warned that there may be some companies that, in the rush to jump on the AI bandwagon, aren’t creating products that may actually benefit brands. “A lot of people are saying they’re doing it, but they are hammers looking for nails,” he said.
ThredUp’s Reinhart said that not every company would have a use for the same kind of AI search model that the company debuted. It allows for queries that are descriptive in ways that don’t relate to the product itself — like “St. Patrick’s Day,” “Buffalo Bills” or “date night outfit.” But ThredUp has millions of unique items on its platform, and it doesn’t do traditional product-driven advertising. The custom tool, he said, wouldn’t necessarily provide much benefit to brands that have a more limited assortment, nor would an out-of-the-box solution work for ThredUp.
“ThredUp has thousands and thousands of new products every hour,” he said. “We don’t go out and shoot model photography in the Atacama Desert and try and convince people [to buy]. And so, our business benefits tremendously from the generative AI piece in the search.”
Yet many Shoptalk attendees were at least curious about what’s out there. Bruce Smith, founder of connected rowing machine Hydrow, headed to Shoptalk from Boston seeking out examples of generative AI tools. He attended sessions, and accepted invitations to vendor dinners that he otherwise would “avoid like the plague,” as he wanted to learn more about the space.
While an exciting advancement, Smith also said that he’s certain human interaction will continue to play a role in the world of retail.
“On the one hand we’re so excited about all of this gen AI and the impact it’s having on our digital lives,” he said. “But we’re hauling our asses to Las Vegas to exchange information in person,” he said, “It highlights how critical that human element is, and that’s not replaceable.”