AI retail checkout: retailers seek to retain the sale

AI retail checkout is emerging as a key fault line for merchants: retailers are adapting product information for chatbot discovery while seeking to keep the final purchase on their own websites. AI-referred visitors appear valuable, yet reported agent-led order volumes remain small at some platforms and retailers remain focused on retaining the data and customer relationship that come with direct transactions.

Adobe Analytics data cited by Reuters showed that visitors referred by AI services generated 41% more revenue per visit than shoppers arriving through traditional channels in June. Separately, Adobe data reported by PYMNTS found that AI-referred traffic to US retail sites rose 393% year on year in the first quarter, based on direct online transactions across more than 1 trillion visits.

The figures measure referral traffic rather than a chatbot or AI agent independently completing a purchase. In this developing model, an assistant can help a consumer find a product, while the merchant or marketplace may receive the shopper on its own site to complete the transaction.

Why do retailers want AI-discovered shoppers to use their own checkout?

A purchase on a retailer’s site gives the merchant a direct relationship with the customer and access to browsing, basket and purchase information that can support later recommendations and loyalty programmes. Reuters reported that retailers including Walmart, Ulta Beauty and Wayfair were updating websites for visibility in chatbot results while trying to retain purchases on their own sites.

Ulta is working with Google to incorporate shopping carts and its Ulta Beauty Rewards programme into AI-powered shopping in Gemini. Josh Friedman, Ulta’s head of digital and e-commerce, told Reuters that the company would prefer shoppers to complete transactions on Ulta’s website. He said visitors arriving through Gemini and ChatGPT showed “double the conversion and intent,” an executive characterisation without a disclosed metric definition.

Etsy offers another example. Rafe Colburn, its chief product and technology officer, told Reuters that people who discover items through ChatGPT typically return to Etsy’s site to buy. OpenAI ended its Instant Checkout tool in March and was then focused on product discovery and helping merchants and marketplaces, including Etsy, use their own checkout systems, Reuters reported.

High-value referrals, limited agent orders

The strength of AI referral traffic does not mean autonomous purchasing is widespread. Etsy Chief Executive Kruti Patel Goyal said on the company’s August 6 second-quarter earnings call that traffic from AI-agent platforms was below 1% of total traffic, although it converted at a higher rate and had a larger average order size. DoorDash Chief Executive Tony Xu said a day earlier that agentic order volume from AI partners remained low across the board.

Shopify reported a faster expansion in the second quarter: AI-referred traffic and orders to its merchants each tripled from a year earlier. President Harley Finkelstein said new-buyer orders from AI channels were arriving at nearly twice the rate of other channels. Shopify also said half of AI-referred sessions reached a product-description page directly, 2.5 times the traditional-search rate.

There are operational constraints as well as commercial ones. Xu said an AI shopping assistant must ultimately hand off tasks including stock checks, routing an order to the appropriate merchant and loading a delivery bag, and described that handoff as unresolved across the industry.

Consumer research points to uneven acceptance of transaction delegation. PYMNTS Intelligence said 48% of online shoppers used AI to research their most recent purchase in one survey. In a separate report, it found that 49% of consumers would allow an AI agent to complete a purchase; among people who did not already use AI regularly, trust in AI platforms to handle a transaction was 3%.

This story draws on original reporting from PYMNTS.