Key Takeaways
- A fundamental split over the definition of agentic commerce -- full autonomy vs. incremental automation -- surfaced at IAB Connected Commerce 2026
- AI referral traffic is boosting retail site growth rather than cannibalizing it, with higher engagement and conversion rates
- AI agents eliminate the "moment of indecision" where ad clicks happen, threatening to reshape retail media's revenue model
The Definition Debate That Divides the Industry
Industry experts debate the definition of agentic commerce at IAB Connected Commerce.
www.emarketer.com"My definition of agentic commerce is the entire shopping process, from product discovery to price negotiation to checkout, all of it done autonomously on your behalf. Anything requiring human approval is simply an evolution of search." Ryan Verklin, retail media and paid media senior lead at Bayer, drew a sharp line at IAB Connected Commerce in April 2026.
Andrew Lipsman, founder and chief analyst at Colosseum Strategy, took a similar stance. "What I don't think it can be is just simply throwing everything that is AI assisted or AI Search Driven, and calling that agentic," he said. "We had AI before we had agentic. Agentic has to be differentiated, and to me it comes down to that level of autonomy." This echoes the position he has consistently outlined in his contributions to The Drum.
Scott Collins, senior director of product strategy at Moloco, offered a contrasting view. "Agentic at its core is that it's taking an action. That can be a sequence of actions, but it doesn't necessarily have to be all of them strung together. There's a spectrum of agentic behavior." His position stands in clear tension with the all-or-nothing framing of Verklin and Lipsman.
This may appear semantic, but as EMARKETER analyst Sarah Marzano pointed out, "We're asking retailers and brands to prepare for a certain future." If the future differs depending on who you ask, so does the strategy. Those who define agentic commerce broadly tend to argue that autonomous purchasing is imminent. Those who define it narrowly see a longer runway. This gap in perception is creating strategic misalignment across the industry.
The Paradox: AI Referral Traffic Is Strengthening Retail Media
A common fear in the agentic commerce discourse is that AI assistants will cannibalize retail media revenue. Early data presented at IAB Connected Commerce told a very different story.
According to SimilarWeb data cited at the event, consumers arriving at retail sites via AI platforms spend more time on site, view more pages, and convert at higher rates than average visitors. "If the referral traffic from AI assistants ends up seeing more of the retailer's real estate and having higher intent, that can only mean good things for retail media," said Marzano.
Lipsman's analysis of the top 15 retailers added further evidence. He found a positive correlation between the share of traffic from generative AI referrals and year-over-year website growth. "If it were true that AI was pulling from their traffic, you'd expect to see a negative relationship," he noted. "Instead, the opposite."
The Walmart case is particularly telling. Marzano reported that 50% of traffic from ChatGPT to Walmart represents net new customers, demonstrating that AI platforms expand retailer reach rather than simply redirecting existing demand. EMARKETER estimates that US ecommerce sales via AI platforms will reach $20.9 billion in 2026, nearly quadrupling from 2025.
When "Indecision" Disappears: A Structural Shift in Retail Media Economics
Even if AI does not steal retail media traffic, a deeper transformation is underway. Collins articulated it precisely: "What agents do is they sort of eliminate that point of indecision. And that point of indecision is where a lot of ad clicks happen on a retail media site."
In the traditional purchase journey, when a consumer searches for "running shoes," no one knows whether they need gym shoes or marathon trainers, or whether they have brand preferences. Within that ambiguity, consumers browse multiple product pages, click sponsored ads, and deliberate. As The Drum reported, retail media's revenue model is fundamentally built on monetizing consumer indecision.
AI agents compress this journey. Through conversation, they clarify intent, read latent signals like hovering, comparing, and cart abandonment, and narrow the options. Rather than nine wasted clicks and one converting click, the model shifts toward CPA (cost-per-action) pricing.
This does not spell the end of retail media, however. Discussion at IAB Connected Commerce 2026 suggested that retailers who build on-site AI agents powered by proprietary data can maintain their influence. "If a retailer that has tons and tons of experience can actually train a model to serve that experience, then there's a reason for that shopper to continue to come to that retailer," said Collins.
The Risks of "Advancing Without a Definition" and the Value of Physical Stores
The risk of moving forward without a shared definition goes beyond scattered investment. At IAB, data was presented showing that over 50% of internet traffic is now non-human, raising fundamental questions about trust when machines act on consumers' behalf.
Amid this uncertainty, physical retail channels attracted renewed attention. Marzano noted that "most of the long tail is just getting started in-store. Retail media is a huge opportunity that is much less under threat by digital discovery paths." Dollar General's AI audio rollout and Grocery TV's in-store media initiatives are examples of physical retail media networks serving as a hedge against digital volatility.
The definition debate ultimately comes down to where we draw the line on AI autonomy. And that line will inevitably differ by industry, product category, and purchase frequency. Auto-replenishment of household essentials and considered purchases of expensive electronics demand entirely different levels of autonomy. Perhaps the quest for a single definition is itself part of the problem.
Summary
IAB Connected Commerce 2026 revealed that agentic commerce is already reshaping retail media's structure, even though the industry cannot agree on what it actually is.
The full-autonomy camp and the incremental-automation camp diverge on strategy. Yet both share two recognitions: AI referral traffic is currently strengthening retail media, and the compression of the purchase journey poses a medium-term risk to existing ad models.
With AI-driven US ecommerce sales projected at $20.9 billion in 2026 -- nearly four times the 2025 figure -- the market is expanding regardless of definitional consensus. What retailers and brands need is not a bet on one definition, but flexible infrastructure that can adapt to multiple scenarios.
