TL;DR: The New Hierarchy of Discovery

• The AI Mall Map: OpenAI’s new Shopping Research feature is the “first entrance to the AI Shopping Mall.” If your brand is not “legible, structured, and machine readable,” you will not appear on its map.
• AI Search is the New Google: Shoppers are skipping traditional search, paid media, and SEO stacks entirely. Instead, they are starting their product discovery with a conversation in ChatGPT.
• The Revlon Warning: Our internal AI visibility tests showed that major heritage brands like Revlon, despite decades of brand equity, disappear in early-funnel AI discovery.
• What Wins Now: AI rewards are not backlinks or domain authority, but structured data clarity, clear review sentiment and distribution, and off-site signals like creator mentions and Reddit threads.

The Shopping Mall Era Is Ending and the New Front Door Just Opened

There have only been a few true revolutions in how people shop. The invention of the mall, the rise of big-box retailers, the dominance of Amazon, and the era of Google Shopping. Each one dramatically shifted where the journey begins.

Yesterday, OpenAI shifted it again.

They released Shopping Research and with it, the first entrance to the AI Shopping Mall. For the first time, shoppers are not starting with a website or a retailer. They are starting with a conversation.

If the mall centralized stores and Amazon centralized transactions, ChatGPT is centralizing decisions. This is the new front door of commerce, and it fundamentally changes how products are found.

Why This Moment Is Bigger Than the Feature

In 2010, the commerce world learned one rule that defined the last decade: Whoever controls the beginning of a shopping journey controls the money. Search dollars, affiliate ecosystems, and brand spend were all built on this idea.

OpenAI now owns the beginning.

Shopping Research lets you talk to ChatGPT the way you would a knowledgeable salesperson. You simply describe what you’re looking for in plain language.

Instead of a static list, the model asks follow-up questions that clarify your intent. It gathers information from specs, availability data, and reviews. It converts all of that into a personalized buyer’s guide that reflects your actual needs.

The experience feels less like browsing a website and more like having a delegated AI agent walk the aisles for you. OpenAI is making this available to all users with almost unlimited usage, which means this high-intent behavior will spread quickly.

This is not ranking. This is not search results. This is guided discovery.

The Product Is Early, but the Shift Is Immediate

The UI is basic, the UX is incomplete, and coverage gaps exist. OpenAI literally says this is only the first version. None of this slows the shift.

The behavioral pattern is already different:

• Personal Context: People are giving the model far more personal context than they ever provided to a search bar. They describe specific use cases, constraints, and preferences in a way search interfaces never capture.
• Deeper Intent: The model pushes them to clarify what matters, forcing a deeper understanding of the actual need behind the request.
• Long-Tail Needs: Queries now stretch into “long-tail” needs that traditional search could never surface. Instead of typing “best vacuum,” they explain their flooring, their pets, their noise sensitivity, and their willingness to maintain the product.
• Conversational Evaluation: People evaluate products inside a conversational frame. It feels like being guided through a decision with someone who is trying to understand them, not sell to them.

Users are not filtering. They are talking. And the system is learning what they actually mean, not just what they typed. This is a new mode of shopping.

The AI Product Feed is Your New SEO

The “Explosion of Citations” is the loudest warning. In some of our experiments, the model used 91 pages of citations to generate a single buyer’s guide. This matters because the model only cites what it can parse.

Our analysis of Revlon proved the cost of poor data: the brand disappeared in early discovery because the model couldn’t properly parse its site or find the creator mentions and Reddit threads it weights heavily.

• The Visibility Crisis (Problem):
  • AI needs clarity and structured data to synthesize an answer.
  • Vague product descriptions are skipped.
  • AI prioritizes trust and tone from outside sources.
• The AI-Ready Fix (Solution):
  • Implement product schema and utilize the Agentic Commerce Protocol (ACP) required fields.
  • Product descriptions must be up to 5,000 characters of actual value—specific details that AI can quote, not vague marketing copy.
  • Include product_review_count and product_review_rating in your feed, as AI trusts social proof more than your own copy.

This isn’t about traditional SEO. It’s about being comprehensible to the machine. The AI Product Feed is the non-negotiable data structure required to move from invisible to recommended.

Why Heritage Brands Cannot Sit Out This Shift

The risk is immediate and visible. Our AI visibility tests already made the risk obvious.

• When we tested Revlon in edition 11, its Overall Discoverability Score was 50.7/100.
• The real problem was the Discovery (early funnel) score of 38.4/100.
• Ask an open-ended question like “best long-wear lipstick” or “clean beauty brands to try,” and Revlon disappears.

AI is now the first opinion customers hear. If you are not present there, you lose the journey before it begins.

The Bottom Line: The Commerce OS Just Shipped

The Shopping Mall Era was about foot traffic. The AI Mall Era is about question traffic.

OpenAI just moved the beginning of the shopping journey into an intent-driven conversational engine that already acts more personally and contextually than anything consumers have used before. The UI can be rebuilt, but the direction is set.

Shoppers will start their journey by asking an AI. And whoever controls the beginning of a shopping journey controls everything that follows.

Brands that are not legible, structured, and machine-readable will simply not exist inside this new discovery economy.

— Tomer