Retail AI Report: The Trillion-Dollar Transition

McKinsey and ICSC Put a Number on the Agentic Commerce Opportunity- and It's $1 Trillion

In what may be the most consequential data release in retail strategy this year, ICSC and McKinsey published Shopping in the Age of AI: Redefining Stores for a New Era, a report built on a survey of 3,004 U.S. consumers that estimates agentic commerce could generate up to $1 trillion in U.S. B2C retail revenue by 2030, with global projections reaching $3–$5 trillion. The stakes of that number extend well beyond the forecasting exercise: it defines a structural reordering in which AI agents, not consumers, become the primary interface between brands and the purchase moment. The report's most operationally urgent conclusion is that physical stores must commit to one of two distinct identities,  convenience-optimized (fast, frictionless, transactional) or discovery-led (experiential, inspirational, irreplaceable by a screen), because the stores caught in the middle, trying to be both, will be the ones most easily bypassed by an AI that can simply order online instead. For C-suite teams still treating AI as a technology investment rather than a strategic positioning decision, this report is the clearest signal yet that the window to define your store's role in an AI-mediated world is closing.

Why This Matters

Agentic commerce shifts the competitive battleground from consumer preference to AI preference. If an agent is making the purchase decision, your store's value proposition needs to be legible to a machine, which means clear product data, competitive pricing signals, and unambiguous positioning. Retailers without a defined identity between convenience and discovery don't just lose customers; they get filtered out entirely.

Source: Business Wire / McKinsey & Company / ICSC

Adobe's Data Makes the Case That AI Traffic Is Now Retail's Best Customer

Adobe Digital Insights released Q1 2026 data this week that deserves a place in every CMO's next budget conversation: AI-driven traffic to U.S. retail websites grew 393% year-over-year, and visitors arriving from AI sources converted into purchases at a rate 42% higher than those arriving from paid search or email- channels that currently command the majority of most retailers' acquisition budgets. The behavioral profile of these visitors compounds the story: they spent 48% longer on site and browsed 13% more pages per session, suggesting they arrive with stronger purchase intent and genuine product interest rather than impulsive click-through. The structural implication is that AI tools like ChatGPT and Perplexity are functioning as a pre-qualification layer, doing the comparison work before the shopper ever lands on a retailer's site, which is precisely why conversion rates are so much higher. What this demands from marketing teams is not incremental adjustment but a reallocation of strategic attention toward what some are calling generative engine optimization (GEO): ensuring that your products, reviews, brand narrative, and structured data are optimized to appear accurately and favorably in AI-generated recommendations, not just in Google's ten blue links.

Why This Matters:

A 42% conversion premium is the kind of gap that rewrites channel allocation decisions. The implication for marketing budgets is direct: if AI-referred visitors convert better, stay longer, and browse more deeply, the ROI calculus on paid search versus GEO investment is already shifting. CMOs who treat this as a trend to monitor rather than a budget decision to make are running a year behind.

Source: Crescendo.ai / Adobe Digital Insights

Gen Z Has Already Moved On from Google, and Your Brand May Not Be Following

The most quietly alarming data point of the week came not from a technology company but from the World Retail Congress in Berlin, where WPP's brand analytics chairman David Roth presented research showing that 17% of all shoppers now begin product searches on ChatGPT or Claude rather than Google,  and among Gen Z consumers, that figure reaches 47%. McKinsey-cited projections at the same conference estimated that AI-driven discovery could represent $750 billion in retail revenue by 2028. Roth's framing was pointed and worth absorbing: every AI has an opinion of your brand today. That observation exposes a strategic blind spot most organizations haven't yet addressed, large language models form their representations of brands from training data that includes reviews, press coverage, social content, and retailer descriptions, none of which were written with AI summarization in mind. A brand that has spent a decade optimizing its Google search presence may be effectively invisible, misrepresented, or outcompeted in the AI discovery layer where its next generation of customers has already migrated. The implication for CMOs is not hypothetical: paid search ROI is quietly eroding at the generational cohort most likely to drive retail growth over the next decade.

Why This Matters

Brand representation in AI is not something you can buy your way into the way you can with paid search. It is earned through the quality, consistency, and structure of everything published about your brand across the web. Most retailers have no systematic approach to auditing what an AI says about them, and the 47% of Gen Z using AI for discovery are forming opinions based on whatever the models learned before the brand team got involved.

Source: WWD / World Retail Congress

OpenAI, Google, and Anthropic Are Building Very Different Versions of Agentic Commerce

A detailed analysis from Digital Commerce 360 this week illuminated something that most retail technology discussions flatten: the three dominant AI platforms are not building toward the same agentic commerce model, and the strategic implications for retailers diverge significantly depending on which infrastructure they integrate with. OpenAI has quietly walked back its in-ChatGPT native checkout ambitions after Walmart found that conversion rates were three times lower for native checkout compared to redirecting users to the retailer's own site — a finding that changes the economics of the entire channel. OpenAI is now focusing on the ChatGPT app ecosystem and deeper Shopify integrations. Google, meanwhile, is deploying its Universal Commerce Protocol to enable native checkout across AI Mode in Search and Gemini, betting that its proximity to existing shopping behavior gives it a structural advantage. Anthropic is pursuing the most architecturally ambitious approach: its Project Deal experiment orchestrated 186 real marketplace transactions using AI agents representing both buyers and sellers simultaneously — a genuine multi-agent commerce model that has no parallel in traditional e-commerce infrastructure. For retail technology executives, the practical takeaway is that platform-neutral architecture matters more than ever, and aligning your commerce infrastructure too tightly with any single AI provider right now carries meaningful platform risk as the competitive landscape continues to shift.

Why This Matters

Walmart's finding that native AI checkout underperformed redirect checkout by 3x is one of the most important data points in commerce this quarter — and most retailers haven't seen it yet. Before committing engineering resources to any single platform's checkout integration, retail technology leaders should pressure-test the assumption that meeting customers inside AI environments improves conversion. The data so far suggests it does not, at least not yet.

Source: Digital Commerce 360

Modern Retail's Investigation Reveals a Widening Gap Between AI App Launches and Consumer Adoption

While brands race to claim territory in the emerging AI app ecosystems of ChatGPT and Claude, a rigorous investigation from Modern Retail this week found that shopper adoption remains, for now, largely theoretical. Roughly 900 apps now exist on ChatGPT, with approximately 10% targeting retail and shopping, including launches from Sephora and Starbucks, while Anthropic's Claude launched its own connector ecosystem with Instacart among the founding partners. The concern is that no brand has yet been able to point to one as a meaningful, measurable revenue driver. Industry analysts quoted in the piece drew direct comparisons to the metaverse and NFT retail experiments of 2021–2022, periods when brands poured resources into platforms that attracted press coverage but not customers. The important nuance the piece surfaces is that the same AI platforms driving that explosive Adobe conversion data are also where these shopping apps live, which means the opportunity is real, but the interface through which customers access it may not be a branded app at all. The smarter bet, several experts argued, is less about building a destination inside ChatGPT and more about ensuring your products and brand appear accurately when AI systems respond to shopping queries, the GEO priority rather than the app-launch priority.

Why This Matters

The metaverse comparison is a useful caution, but it shouldn't be read as a reason to disengage from AI platforms entirely. The distinction is between presence and destination: being findable and accurately represented inside AI environments is a fundamentally different investment than building a branded app few people will open. The former is a structural necessity; the latter is discretionary, and right now the returns aren't there.

Source: Modern Retail

Dollar General's 12,000-Store AI Audio Network Is a Case Study in Physical Retail Media Scale

While much of this week's news focused on digital AI channels, Dollar General quietly executed one of the most operationally significant AI deployments in brick-and-mortar retail this quarter: the expansion of its AI-enabled in-store audio network, powered by QSIC, to 12,000 locations across 48 states by Q2 2026. The platform is not simply a music curation service; it integrates point-of-sale data to generate localized, real-time audio advertisements and closed-loop measurement that links ad exposure directly to in-store purchase outcomes. For Bank of America Securities analysts, who flagged the rollout as a margin improvement lever, the significance is that Dollar General is converting its existing store traffic, a sunk cost asset, into a measurable, addressable media network with first-party sales data attached. This is the physical-world equivalent of what Amazon built with its advertising business: a high-margin revenue stream layered on top of infrastructure that already exists. For large-format retailers watching Amazon's ad revenue growth with envy, Dollar General's model offers a practical, technology-enabled roadmap that does not require a digital marketplace to execute.

Why This Matters

Dollar General's move reframes what store infrastructure is worth. Any retailer operating at scale has a media asset inside its stores,  foot traffic, dwell time, captive attention, that is currently going unmonetized. The QSIC integration shows that the technology to close that loop between ad exposure and purchase is now commercially available and deployable at 12,000-location scale. Retailers with strong store networks and thin margins should take this model seriously as both a revenue line and a margin lever.

Source: Retail Dive

What to Watch

IBM Think 2026 runs through May 7 in Boston, and additional retail-specific announcements from the event, particularly around agentic supply chain AI and autonomous inventory management, are expected in the coming days; COOs and Chief Merchandising Officers should monitor these closely as the deployments move from concept to vendor roadmap. The generative engine optimization space is moving fast enough that the gap between early movers and laggards is already measurable in Adobe's conversion data, expect the first GEO-focused agency practices and auditing tools to become widely available before Q3. And as Google's AI Mode in Search continues its rollout, watch closely for first-party data on how native AI checkout performs at scale, that single metric may be the most important number in retail commerce for the remainder of 2026.