Agentic Commerce $1 Trillion by 2030: Is Your Catalog Ready?

• Fact: McKinsey and ICSC forecast $1 trillion in US retail revenue to be orchestrated by AI agents by 2030 — rising to $5 trillion globally.

• Impact: 68% of consumers already used at least one AI tool in their last three shopping interactions; the adoption curve is steeper than most brand plans assumed.

• Surprise: The real bottleneck is not agent capability — it is catalog quality. Brands that win will be those whose product data is machine-readable, not just marketing-friendly.

Four years is a very short runway for a trillion dollars. That, in essence, is what a new joint report from ICSC and McKinsey & Company is telling brand managers and COOs who still have “agentic commerce” parked under Future Initiatives on their roadmap. The number — $1 trillion in orchestrated US retail revenue driven by AI agents by 2030 — lands with the kind of specificity that forces a different conversation inside the boardroom.

What’s striking about this moment is not the size of the forecast alone, but the speed of adoption already underway. McKinsey finds that 30–45% of US consumers are already using generative AI for product research and comparison. Sixty-eight percent used at least one AI tool in their last three months of shopping. We are not talking about a pilot program at a handful of tech-forward retailers. This is mainstream consumer behaviour shifting in real time.

What “Orchestrated Revenue” Actually Means

The language in the RetailDive coverage of the ICSC/McKinsey report deserves unpacking. “Orchestrated” revenue is not simply revenue influenced by AI recommendations — it is revenue where an autonomous agent browsed, compared, negotiated terms, and completed a transaction with minimal or zero human intervention. Think Amazon’s Buy for Me, Visa’s agentic payment rails, or OpenAI’s emerging merchant integrations.

The distinction matters enormously for how brands should respond. Influencing a human consumer who reads your copy is one capability set. Getting selected — or rejected — by an AI agent evaluating thousands of options in milliseconds against structured criteria is an entirely different problem. The agent doesn’t read your headline. It queries your attributes.

Other forecasters are somewhat more conservative but point in the same direction: Bain estimates agentic commerce reaches $300–500 billion in the US by 2030, roughly 15–25% of all e-commerce. Morgan Stanley pegs a narrower $190–385 billion. Even at the low end, these are not rounding errors. They are channel-defining numbers.

The Catalog Problem Hiding in Plain Sight

Here is the awkward truth for most brands: the shift to agentic commerce is not primarily a technology problem. It is a data problem. AI agents rely on structured attributes — material composition, compatibility, usage scenarios, dimensions, certifications — as their primary filters. Marketing copy, brand storytelling, and lifestyle photography are essentially invisible to an agent evaluating a purchase against a user’s stated intent.

Epinium data

In our work optimizing product catalogs across 500+ brands on 12 Amazon marketplaces, Epinium consistently finds that listings with complete structured attributes — technical specs, usage scenarios, and compatibility data — are surfaced by AI shopping agents at roughly three times the rate of listings that rely on unstructured marketing copy alone. Catalog completeness has become the new SEO.

What we’re seeing at Epinium is that the brands scrambling hardest right now are not those without AI strategies — they are those with excellent brand marketing and poor attribute data. They spent years perfecting copy that speaks to humans, and now the first buyer to evaluate their listings is often a machine.

The good news is that this is fixable, and faster than most teams assume, provided the right infrastructure is in place. We examined recently how Amazon’s Rufus agent — which grew 115% in query volume — surfaces products differently from traditional keyword search, and the pattern was consistent: attribute density wins over creative copy every time.

Is your product catalog ready for AI agents? Epinium’s catalog management platform structures your product data for both human and machine discovery →

From Forecast to Factory Floor: What Early Movers Are Doing

Two retailers announced concrete moves this week that show where smart money is going. Wayfair’s CEO Niraj Shah said the company wants to “be everywhere” in agentic AI — both through advertising integrations and by making its inventory directly callable by external agents. Etsy, meanwhile, launched its native app inside ChatGPT, letting users browse handmade goods through a conversational interface and complete purchases without ever leaving the chat. Neither company is waiting for 2030.

For brands selling through these platforms, the implication is immediate: the products that surface inside agent experiences will not be chosen by a merchandising team. They will be selected algorithmically, against structured data. The catalog readiness challenge is here today, not in the future tense the headline suggests.

FAQ

What exactly is agentic commerce, and how is it different from AI-assisted search?

Agentic commerce is where an AI agent autonomously browses product options, compares prices and attributes, and completes a purchase on behalf of a user with minimal human intervention. AI-assisted search still requires a human to read results and click buy. In agentic commerce, the agent does all of that — including the transaction — often before the user sees a confirmation screen.

Does the $1 trillion forecast include B2B purchases?

The McKinsey/ICSC $1 trillion figure focuses on US business-to-consumer retail. McKinsey’s broader global estimate of $3–5 trillion likely incorporates B2B procurement flows, where agentic purchasing has arguably faster adoption potential given existing API integrations and structured supplier catalogues.

What should a brand do in the next 90 days to prepare for this shift?

Start with a structured attribute audit of your top 20% of SKUs by revenue. Specifically check technical and functional specs, not just marketing fields — material, dimensions, compatibility, certifications. Then verify that data reaches each marketplace in structured format rather than buried inside long-form description copy. Pricing and promotional data should be separately structured and machine-readable.

Is this only relevant for large brands with dedicated data teams?

Mid-market brands may feel the impact faster. Large brands typically have dedicated catalog teams already working on this. Mid-market brands selling through Amazon, Wayfair, or Etsy will find their products either appear in agent-driven surfaces or they don’t — and the revenue gap will show up before they understand the cause.

When does a “wait and see” posture become a serious competitive liability?

Arguably it already has. The 30–45% of consumers already using generative AI for product research are conducting proto-agentic searches today. Acting after agent commerce is fully normalised means rebuilding a catalog while competitors have had structured data in place for two or three years. The compounding effect of that head-start is significant.

The trillion-dollar forecast will look either prescient or conservative depending on how quickly the agent-payment layer matures. What won’t change is the underlying mechanic: machines evaluating products against structured data, at speed, at scale. The brands that invest in catalog quality now are building a durable advantage, not a temporary edge.

Ready to make your catalog agent-ready? Epinium’s platform helps brands structure, enrich, and distribute product data across the marketplaces where AI agents shop. Discover how Epinium prepares your catalog for the agentic commerce era →