E-Commerce Retailers Are Betting Big on Agentic Commerce, Study Finds
Almost all enterprise e-commerce retailers have already invested in AI capabilities, and nearly half plan to spend $1 million or more on agentic commerce in the next 12 months. That is the headline finding from a LogicBroker survey of over 600 enterprise e-commerce leaders released March 18, and it signals how quickly AI has moved from back-office experiment to front-line business priority.
How Widespread AI Adoption Already Is
Nearly 96% of e-commerce retailers surveyed have invested in AI capabilities in some form. The primary motivations are revenue growth and improving the customer experience, followed by cost reduction and operational efficiency.
The shift is significant. A year ago, the dominant model was deploying AI in the background while keeping humans in the loop for anything customer-facing. That model is giving way to one where AI interacts directly with shoppers at multiple touchpoints.
“Keeping humans in the loop with AI is still sound advice, but the application has shifted,” said Julie Geller, principal research director at Info-Tech Research Group. “The question isn't about whether to keep humans involved; it is where in the customer journey that involvement matters most and how you design for it.”
Where Retailers Are Focusing Their AI Spending
The top three applications are all customer-facing. AI-powered product discovery, AI chatbots, and personalized recommendations lead the list of investments. On the operational side, 44% are investing in pricing optimization and 43% in automated inventory management.
The customer-facing focus reflects where retailers see the clearest revenue opportunity. Product discovery in particular has drawn attention as AI tools have shown they read behavioral signals more accurately than traditional keyword search. As Geller put it: “A shopper who looks at something three times and never buys is communicating something, and AI reads that better than any keyword logic.”
Pricing optimization and inventory management have also proven reliable applications because the variables are easier to quantify and mistakes do not damage a customer relationship directly.
Where AI Chatbots Are Still Falling Short
Despite the investment, the customer experience delivered by AI chatbots remains inconsistent. Shoppers value the 24/7 availability and fast response times, but AI chatbots consistently struggle with complex inquiries. Customers frequently describe them as a barrier to reaching a human rather than a genuine source of help.
Geller's assessment is direct: “Many of these agents sit in an uncomfortable middle ground, neither convincingly human nor comfortably machine. Responses can be hollow, nonsensical, or arrive too late to matter, and either way, the experience actively damages the brand it was meant to serve.”
The data reflects this tension. Two-thirds of consumers still prefer talking to a human, according to a YouGov survey conducted on behalf of Pegasystems. That preference is unlikely to shift until the handoff between AI and human agents improves.
The handoff is where most companies are currently underinvesting, according to Geller. When a customer transfers from an AI chatbot to a live agent, they should never need to repeat themselves. The agent should arrive fully informed and ready to continue the conversation. “Right now it rarely is,” Geller said.
The Cost Question
The ROI of AI is not as straightforward as the investment appetite suggests. Gartner predicts that costs per resolution for generative AI will exceed $3 by 2030, which is more expensive than many offshore agents. Rising data center costs, AI vendors shifting from subsidized growth toward profitability, and increasingly complex use cases are all contributing to upward cost pressure.
That does not mean retailers should pull back on AI spending. It does mean that the strongest returns come from deploying AI where it operates in the background, handling quantifiable tasks with immediate feedback loops rather than taking on customer interactions where the cost of a poor experience is measured in lost trust.
What This Means for the Broader Market
The LogicBroker findings land in the middle of a broader reset in agentic commerce expectations. OpenAI's recent retreat from Instant Checkout demonstrated that even with a massive user base, converting AI-driven product discovery into direct transactions is harder than anticipated. Retailers are drawing a similar lesson from the chatbot data: reach and scale do not automatically translate into a better customer experience.
The retailers most likely to see returns on their agentic commerce investments are those treating AI as infrastructure rather than a customer service shortcut. Product discovery, pricing optimization, and inventory management all fit that model. Customer-facing chatbots that replace human agents before the handoff experience is built properly do not.
