The year is 2022.
A growing consumer brand has a great product at a competitive price. Its product data, however, is less impressive. Descriptions vary across retailers. Attributes are inconsistent. The company knows there’s room for improvement, but other priorities keep getting in the way.
Search for the product category online, and the brand appears near the bottom of Page 1. Maybe Page 2.
Not ideal. But that great product still has a fighting chance of making it into a shopper’s cart. There is hope.
The year is 2027.
The product is still great. The price is still competitive. But the online shopper no longer scrolls through pages of search results. Instead, they tell an AI shopping agent what they want and then let it discover, compare, and select a product.
This time, the brand never gets its chance.
Incomplete or inconsistent product data has rendered that great product invisible online. The agent can’t identify and match it to the shopper’s request, so it chooses something else.
The sale is lost, quickly and quietly. So are the repeat purchases that might have followed.
This scenario may reflect the future of online shopping, and the transition is underway. AI agents influence product discovery and send shoppers to retail and brand websites, even as autonomous purchasing remains in the early stages.
“AI is going to return maybe three things, where a search engine would give you 20,” said Nick Latwis, director of innovation at GS1 US, describing one example of the difference.
That changes the stakes for brands. As discovery shifts from search-based to AI-driven, visibility will depend on the AI agent’s ability to identify, understand, and consider the product.
The Situation: Transition from Ranking to Selection
For decades, digital commerce has been a competition for top search rankings, with brands investing heavily in search engine optimization.
Agentic commerce shifts the dynamic.
Instead of endless pages of search results — with the first two or three pages most likely to draw consideration — an AI agent may instantly search more broadly than the average shopper ever would, though not necessarily as deeply. Ultimately, it will present a select few choices.
The efficiency appeals to the shopper.
And it challenges the brand.
“If you ask, ‘Hey, look at all of these 100 different options and give me the top options,’ it might not look [at everything with] the same level of due diligence,” Latwis said.
So the focus has shifted from ranking to selection.
A product that landed on Page 2 of a traditional search may be disadvantaged, but it remains visible. When an AI agent is tasked with presenting the three best options, the shopper may never see option 4 or 5, let alone option 50.
That journey may become even more compressed as agents take on additional responsibilities. Today, Latwis said, AI-driven referrals to brand and retailer websites are far more prevalent than autonomous purchasing. Perhaps sooner rather than later, AI agents will regularly complete purchases with less oversight from the shopper.
That means price, availability, reviews, and brand recognition still matter. But they only enter the equation once a product is in the AI agent’s consideration set.
For brands, that creates a question at the top of the funnel: Can AI confidently identify and understand the products you sell?
The Question: Can AI Understand Your Products?
Human shoppers make connections that product data doesn’t always make explicit, recognizing that slightly different descriptions refer to the same product or filling in missing details from context.
AI agents need more reliable signals.
Brands control information on their own websites, Latwis said, but retailers and third-party marketplaces may alter that information to fit their own systems. Maintaining a consistent product identity helps connect disparate sources.
“If there’s that central grounding identifier, that’s probably a good place to start just to help corral information,” Latwis said.
That’s one role of the Global Trade Item Number (GTIN), the unique identifier assigned to products and commonly encoded in barcodes. GS1 Digital Link takes that further. It expresses the GTIN, along with additional details such as potential allergens, recalls, and other batch or serial number information that both people and machines can use to narrow down their search for the right product.
Identity is only the beginning. Accurate, standardized attributes — descriptions, weights, dimensions, product categories, and more — help agents understand and compare products.
Then consider how products are presented online.
Product pages are typically designed to appeal to human eyes. That’s still important, but now brands must consider a second audience: machines.
Latwis pointed to the GS1 Web Vocabulary as one tool to make online product information more structured and machine-readable. Rather than asking AI to infer product identity based on prose, images, or documents, structured metadata makes those facts easier to extract.
The Consequences: When Good Products Become Invisible
The most dangerous lost sale is the one the brand never knows it lost.
In traditional e-commerce, failure leaves clues. A product ranks poorly. Shoppers visit a page but leave quickly. Abandoned carts pile up.
In those scenarios, brands can see their shortcomings and adjust accordingly. Agentic commerce can erase those cues.
“If the products aren’t surfaced in the first place, a human doesn’t even know to ask those questions,” Latwis said.
Tools are emerging to test how large language models respond to queries about particular products or consumer needs, but Latwis said the analytics ecosystem remains less mature than the tools companies use to measure search performance.
Larger organizations have more resources to optimize product information for both human shoppers and machines. Smaller brands may lack the bandwidth or technical expertise.
This is the nightmare scenario for up-and-coming brands.
An excellent product can be available, competitively priced, and ideally suited to the consumer’s needs, yet lose to a competitor’s product because the AI agent can identify and evaluate it more easily.
No second or third page for rigorous shoppers to find it. The product is simply invisible.
The Preparation: Making Products Agent-Ready
Brands can’t predict which AI agents shoppers use, nor can they know with certainty how the AI market will evolve.
They can only control what they can control.
Start by establishing a consistent identity for each product, reconciling conflicting information across systems, and ensuring critical attributes are accurate and complete. Then consider how easily a machine can consume it.
“Look at your web pages, any PDFs you have embedded, things like that,” Latwis said. “If a human looks at it, good. But how do we optimize this for an AI agent and AI audience, both for accuracy but also for efficiency?”
Latwis also recommends experimenting directly with AI tools to understand how agents find and interpret product information.
Technology leaders can help colleagues understand how agents access and interpret information, while supply chain, data, and other stakeholder teams can establish clear ownership of product information throughout the process.
GS1 US Standards provide a foundation.
Latwis pointed to GTINs as a starting point for consistent product identity. He also recommends that brands look beyond the GTIN to GS1 Digital Link so AI agents have access to richer, brand-authorized product information. GS1 US is also investigating how the GS1 Web Vocabulary can make product metadata on websites more consumable by AI agents.
The year is 2026.
No one knows how AI shopping will evolve. Whatever capabilities, agents, or platforms rise to the top, the brands that strengthen that foundation today can give their products a better chance of being found tomorrow.
It’s the first step toward ensuring visibility in an AI-driven marketplace where invisibility is the enemy. And visibility is the first key to success.
Learn more about how agentic commerce is changing product discovery and what it means for brands.