The storefront isn't dying — it's just invisible to you now, and most merchants still can't see it either.
The Summary
- PayPal's first U.S. Agentic Commerce Pulse Survey reveals 95% of merchants already detect AI agent traffic, but only 20% have machine-readable product catalogs that agents can actually use.
- The gap between agent demand and merchant infrastructure is creating a new digital divide where structured data matters more than scale.
- LLMs privilege the most structured, trustworthy data — not the biggest catalog — opening opportunity for smaller merchants who get their data house in order.
The Signal
We're watching commerce bifurcate in real time. On one side: AI agents crawling product catalogs, making purchase decisions for humans who asked for "running shoes for plantar fasciitis under $150 with next-day shipping." On the other: merchants who built their entire business around human eyeballs and Google's keyword auction, now scrambling to speak a language their storefronts were never designed for.
PayPal's survey of 498 decision-makers quantifies what builders already suspected: the infrastructure gap is massive. Nearly every merchant can see agent traffic. ChatGPT crawlers. Google Gemini bots. Perplexity scrapers. They're all there in the logs. But seeing traffic and serving it are different problems entirely.
"LLMs don't inherently privilege the largest catalog; they privilege the most structured, most trustworthy data signal."
Only one in five merchants have done the foundational work: structured product catalogs in machine-readable formats, interoperable APIs, agent-compatible checkout systems. The rest are running storefronts that might as well be written in a dead language as far as AI agents are concerned. Mike Edmonds, PayPal's VP of agentic commerce, frames it clearly: the competition moved from Google Ads and SEO to how your products appear across LLMs and digital marketplaces.
This isn't a future-state problem. According to PayPal's data, 86-94% of businesses across all segments expect agentic commerce to impact them positively within 12-24 months. Many report it already has. That timeline matters because it means merchants have maybe one product cycle to get this right before their competitors do.
The shift from keyword-driven to intent-driven search changes everything about discovery:
- Consumers aren't typing "running shoes" anymore
- They're asking for highly specific recommendations based on multiple variables
- Agents need structured data to answer those queries accurately
- The merchant with clean data beats the merchant with more SKUs
PayPal CTO Srini Venkatesan's observation about LLM privilege is the key unlock here. Scale doesn't automatically win anymore. A small merchant with impeccable product data — accurate specs, real-time inventory, clear return policies, trustworthy signals — can outrank a massive catalog full of unstructured noise. That's a structural advantage the long tail hasn't had since Web1.
But there's a darker read too. The invisible storefront economy means most purchase decisions happen before a human ever sees your site. The agent has already filtered, compared, and chosen. You're competing in an auction you can't see, optimizing for a customer you never meet, hoping your data schema was clean enough to make the cut.
The Implication
If you're building for commerce, the next 18 months are about infrastructure, not marketing. Getting your product catalog into machine-readable formats isn't a nice-to-have anymore. It's table stakes. And if you're building agents, this is the wedge: merchants need tooling to translate their messy product data into something LLMs can actually use at decision time.
Watch for the picks-and-shovels layer here. Someone's going to build Shopify-for-agents. Clean APIs, structured catalog tools, agent-native checkout. PayPal's clearly positioning for payments in this layer, but the real opportunity is helping merchants cross the infrastructure gap before their competitors do.