AI agents are becoming a major e-commerce channel. Will retailers beat them or join them?

Sponsored by

You can’t predict market shocks. But you can prepare for them. Rithum can help you strengthen your operations with better data accuracy and smarter automation. Streamline your commerce ops.

If you can’t beat them, join them. That’s the approach some retailers are taking with a powerful new middleman in e-commerce: AI-powered agents.

Earlier this month, Walmart announced a partnership with ChatGPT creator OpenAI, which launched its new protocol for funneling customers seamlessly from “chat to checkout.” The announcement was the third in as many weeks, with Shopify and Etsy also striking deals with the AI juggernaut.

Walmart CEO Doug McMillon explained the technology as the inevitable next step in the evolution of e-commerce. First, there was the search bar. Now, there are chatbots and, for now at least, capitalizing on that change means partnering with third parties such as OpenAI.

“We are running towards that more enjoyable and convenient future with Sparky and through partnerships including this important step with OpenAI,” McMillon said.

Yet retailers’ willingness to embrace the new comes as AI agents are already gaining ground among online shoppers. AI referral traffic to retail websites surged 4,700% YoY in July, according to data from Adobe Analytics, and it’s expected to jump another 515%–520% over the holidays.

For AI boosters, the benefits couldn’t be clearer: “Agents are a win-win for both the consumers and the merchants,” Melissa Bridgeford, CEO and co-founder of Wizard, an AI-powered agent specializing in e-commerce, told Retail Brew. “Agents are really going to solve that analysis-paralysis decision, helping consumers not only find what they want, but also make decisions around what they want.”

Control issues: However, integrating with AI agents such as ChatGPT could also be a purely practical decision, aimed at retaining some control over a customer experience that is increasingly being mediated by third-party channels.

“There’s increasing concern that they have less agency in putting that story in front of customers,” Nikki Baird, VP of strategy and product at Aptos, said. “AI shopping is going to go through the roof this holiday season, and it’ll be about price and availability. It won’t be about my value proposition as a brand.”

While more and more web traffic is coming from AI agents, Baird said, many of these shoppers are being funneled directly to product pages, bypassing the category pages where customers would normally stop to browse.

Ya Wen, a former general manager at Amazon and SVP of the Americas at Payoneer, a payment services platform that works with e-commerce sites, said AI agents are just the latest distribution channel to gain purchase in e-commerce, and recent efforts at integration reflect a desire to follow the customer.

“They’re basically saying, ‘We recognize this is where the customers are going, so…we want to drive deeper integration to hopefully have better’—I don’t want to say control—‘but to provide better experience to those search consumers,’” he said.

Down to zero: Keeping up with consumers may require learning how to navigate a radically altered landscape for e-commerce. For instance, some experts believe retailer websites could soon become obsolete.

Bridgeford predicts that within the next decade, traffic to retail websites will fall to effectively zero. “It’s just going to slowly lose market share, and volumes are already dropping, and I think that trend is going to rapidly continue,” she said.

Yet even with the core of their e-commerce models under threat, retailers are beginning to see AI agents “less as a foe and more as a friend,” she added.

Whether that goodwill continues could depend on AI agents continuing to drive sales—one recent study found they still lag behind other channels in terms of conversions. A working paper from the University of Hamburg and the Frankfurt School of Finance & Management found that referrals from organic large language models (oLLMs) such as ChatGPT trailed behind traditional digital channels such as paid search.

The authors wrote that “the widely held expectation that oLLM is already superior to traditional channels is not supported by the data,” though it did show the channel was moving in a positive direction.