The traditional e-commerce funnel is being disrupted before shoppers ever reach a retailer's website, as consumers increasingly turn to artificial intelligence assistants to discover, compare and evaluate products.
Entrepreneur and e-commerce strategist Jose Daniel Duarte Camach o says this shift is creating a new competitive challenge for online businesses: winning the customer may increasingly depend on whether products can be accurately discovered, understood and recommended by AI systems before a conventional website visit takes place.
Recent data points to a rapid change in shopping behavior.
Salesforce's 2026 State of Commerce research found that use of agentic search as the first step in the shopping journey grew 200% year over year. The study combines responses from 3,450 commerce professionals and approximately 4,690 consumers with behavioral data from more than 1.5 billion global shoppers.
Traffic arriving at commerce sites from AI chats increased between 150% and 428% year over year in every quarter measured, while overall site traffic grew only at single- to low-double-digit rates. At the same time, product discovery through traditional search declined 15% between August 2025 and May 2026, while discovery through newer channels including AI assistants, social AI and delivery applications increased 38%.
For Duarte Camacho, those numbers suggest that e-commerce companies need to rethink where the customer journey actually begins.
“For years, online retailers optimized around the assumption that customers would start with Google, a marketplace or the retailer's own website,” said Jose Daniel Duarte Camacho. “AI is breaking that assumption. If the first meaningful conversation about a product happens inside an AI assistant, brands need to think about how their products are represented before the shopper ever reaches the storefront.”
Product Discovery Is Becoming Conversational
Traditional e-commerce discovery has largely revolved around keywords, category pages, filters and search-engine rankings.
AI-assisted shopping introduces a different model.
A consumer can now describe a problem, budget, preference or intended use in natural language and ask an AI system to narrow dozens or hundreds of potential products into a handful of recommendations.
Adobe research published in 2026 found that 86% of surveyed consumers use some form of AI during their retail journeys. Thirty-six percent reported using generative AI chatbots, while 52% used AI-powered product recommendations and the same percentage used image-search capabilities.
The implications extend beyond adding a chatbot to an online store.
Duarte Camacho argues that businesses need to consider whether their product information can be interpreted effectively by AI systems across multiple channels.
“Product data is becoming part of the sales conversation,” Duarte Camacho said. “Descriptions, specifications, availability, pricing, reviews and other structured information need to be accurate and consistent because AI systems increasingly depend on that information to decide which products deserve to appear in a recommendation.”
AI-Referred Shoppers Are Becoming More Valuable
The rapid growth of AI-driven traffic would have limited commercial importance if those visitors rarely purchased.
Evidence now suggests the opposite.
Adobe Digital Insights reported that in March 2026, visitors referred to retail websites by AI sources converted at a rate 42% higher than non-AI traffic. A year earlier, AI-referred visitors had converted at nearly half the rate of other traffic, representing a substantial reversal in shopping behavior.
Adobe also found that 79% of consumers using AI for online shopping said an AI assistant increased their confidence in a purchase, while 69% said they were less likely to return an item purchased with AI assistance.
Duarte Camacho believes those results reflect an important characteristic of conversational product discovery: shoppers may reach retailers with more developed purchase intent.
“When someone has already used AI to compare options, understand specifications and narrow the decision, that visitor may arrive much further down the purchasing journey,” he said. “The strategic question for retailers becomes not only how to attract more traffic, but how to become one of the products that AI considers relevant enough to recommend.”
Retailers Are Racing to Catch Consumer Behavior
Businesses appear to recognize the shift, but adoption remains uneven.
Salesforce found that only 28% of commerce organizations currently use agentic AI, even though another 44% expected to implement it within six months. Meanwhile, 86% of commerce leaders said AI is increasing customer expectations.
Deloitte's 2026 Retail Industry Global Outlook points in the same direction. Nearly 68% of surveyed retail executives expect to deploy agentic AI for significant operational or enterprise activities within the next 12 to 24 months. The report also notes that AI-chat referrals already represent 15% to 20% of total referral traffic for some retailers.
For Duarte Camacho, however, deploying AI should not become the objective itself.
Retailers need to ensure that the infrastructure supporting AI-driven commerce — product catalogs, inventory information, customer data, fulfillment systems and pricing — is reliable enough to support automated interactions.
“An AI agent can create an extremely sophisticated buying experience, but it cannot compensate for inaccurate inventory, inconsistent product data or poor fulfillment,” Duarte Camacho said. “The visible AI experience is only as good as the commerce infrastructure underneath it.”
The Website May No Longer Own the Entire Conversion Journey
Perhaps the most significant development is that discovery and purchasing are beginning to merge outside traditional e-commerce properties.
Deloitte reports that the next generation of online commerce is beginning to integrate product discovery and checkout directly within large language models. Retailers and technology platforms have already begun developing experiences that allow consumers to move from conversation to product selection and purchase with fewer transitions between platforms.
That could fundamentally alter a long-standing e-commerce assumption: that every successful digital journey should ultimately bring the customer onto the retailer's website.
Duarte Camacho believes the emerging model will require brands to think less about controlling every customer touchpoint and more about maintaining a consistent commerce experience wherever transactions originate.
“The storefront is becoming distributed,” he said. “A sale might begin in an AI conversation, move through a payment service and rely on a retailer's fulfillment infrastructure without following the traditional path businesses spent years optimizing.”
Conversion Friction Has Not Disappeared
AI may transform discovery, but retailers still face a familiar problem once customers decide to buy.
Baymard Institute currently places the average global online shopping-cart abandonment rate at 70.19%, based on years of tracking e-commerce checkout behavior.
The figure demonstrates why attracting high-intent shoppers through AI does not eliminate the need for basic e-commerce execution.
Mobile usability, transparent costs, account requirements, payment options, delivery expectations and checkout design can still determine whether intent becomes revenue.
Duarte Camacho says businesses should therefore treat AI discovery and conversion optimization as connected priorities rather than separate initiatives.
“There is little value in building an intelligent discovery experience that sends customers into a frustrating checkout,” he said. “The entire path has to work. AI can increase intent, but the retailer still has to convert that intent into a successful transaction.”
Preparing E-Commerce for Machine-Mediated Shopping
Duarte Camacho expects online businesses to place greater emphasis on several capabilities as AI becomes a more influential commerce intermediary:
- Structured and accurate product data that allows AI systems to interpret products, features and availability correctly.
- Consistent information across channels, reducing conflicts between websites, marketplaces and external AI experiences.
- Real-time inventory and pricing visibility so recommendations reflect what customers can actually purchase.
- Strong product differentiation, giving both consumers and AI systems meaningful reasons to distinguish one offering from another.
- Frictionless checkout and payment experiences capable of converting higher-intent traffic.
- Unified customer and commerce data that supports personalization without fragmenting the customer journey.
- Measurement of AI-originated traffic and revenue, allowing companies to understand how conversational discovery affects acquisition and conversion.
The strategic shift is significant because AI may alter not simply how people shop, but how products compete for consideration.
“In traditional e-commerce, businesses fought for rankings, clicks and website traffic,” Duarte Camacho concluded. “In AI-mediated commerce, they will increasingly compete to become the recommendation. That requires a different way of thinking about product information, technology and digital visibility.”
As conversational search grows and AI-generated recommendations become embedded in consumer behavior, e-commerce businesses face a new challenge: the purchasing decision may increasingly begin before customers encounter the retailer itself.
For companies prepared for that transition, AI-driven discovery represents an opportunity to reach consumers with unusually high intent. For those that are not, the risk is becoming invisible at precisely the point where tomorrow's purchasing decisions are being formed.
About Jose Duarte Camacho
JD Duarte is originally from Heredia, Costa Rica. He has been an entrepreneur and business owner for more than 20 years and divides his time between his existing operations and researching new possibilities in which to invest. When he's not dedicating time to his businesses, he spends time with his supporting wife and two children.
— WebWireID363297 —
