Two-thirds of shoppers aged 18–29 now discover products on TikTok rather than on merchant websites. During Black Friday 2025, social commerce channels drove 21–28% of all transactions.1 These shoppers have already moved on from the traditional storefront, and they aren’t coming back.
The shoppers who do still use search engines aren’t clicking through, either. Nearly 60% of all Google searches in the U.S. now end without a single click, up year-over-year across every measured category.2 Gartner projected in early 2024 that traditional search engine volume would drop 25% by 2026 as AI chatbots absorb queries that once led shoppers to merchant sites.3 Even the most conservative reading of the data points in the same direction.
And now, AI shopping agents have arrived. In 2025 and early 2026, OpenAI launched Operator, Perplexity introduced Instant Buy, Amazon’s Rufus reached 250 million shoppers and drove an estimated $10–12 billion in incremental sales, and Google unveiled the Universal Commerce Protocol at NRF 2026. By Cyber Week 2025, AI agents influenced 20% of all global orders, totaling $67 billion.4 Morgan Stanley projects that agentic commerce will account for $190–385 billion in U.S. e-commerce spending by 2030.5
These agents will never browse your homepage. They prioritize structured, machine-readable product data, and catalogs that don’t provide it become invisible to a channel that Morgan Stanley projects will reach $190–385 billion by 2030.
Athos Commerce is an intelligent discovery platform that helps ecommerce brands connect shoppers to the right products across on-site and off-site channels. As CTO Suhas Gudihal puts it, “In agent commerce, the agent controls the decision. So the agent becomes the gateway, and your data becomes your storefront, not your website anymore.”
Your product data has a visibility problem
The gap between what shoppers expect and what product catalogs deliver is already measurable. In an Athos Commerce/Pixel survey of 801 consumers, 37.8% of shoppers over 60 said they leave a product page when the product information is insufficient. Younger shoppers compensate by checking reviews, watching unboxing videos, and scrolling through social proof. They’ll do the research themselves.
AI agents won’t. They pull from multiple sources, including product pages, third-party reviews, and marketplace listings, but they weight structured, machine-readable attributes above everything else. When key attributes are missing from your feed, agents have to piece together answers from noisy, inconsistent sources or skip your products altogether.
The scale of this gap is significant. According to Gudihal, fewer than 10% of merchants have machine-readable product data today. Most catalogs contain three to four attributes per product, while agents need dozens to match a shopper’s query with confidence.
90% of my customers’ product descriptions miss out on all of those things, because somebody will ask, ‘Is this easily assembled?’ You don’t put that in the description right now.
Suhas Gudihal
CTO of Athos Commerce
Merchants have spent years optimizing storefronts, checkout flows, and paid media. The product data feeding all of those investments remains incomplete. That gap was tolerable when human shoppers could browse category pages, scan images, and fill in the blanks on their own. It’s not tolerable when an agent evaluates your catalog in milliseconds and moves on if the answer isn’t there. The data you show an AI agent isn’t the same data you show a human shopper, and most merchants haven’t built for that distinction yet.
Why your business model needs to change
In traditional ecommerce, the merchant controls the experience. You design the homepage, curate the search results, and merchandise the category pages. The website is the storefront, and everything a shopper sees passes through your hands.
In agentic commerce, that control shifts. The merchant controls the data, and the agent controls the experience. Your product feed becomes the only thing the agent evaluates. A feed with missing attributes, outdated availability, or vague fulfillment terms doesn’t get a second look.
“It is not an incremental upgrade from what we are,” Gudihal says, “but it is a shift from digital storefront architecture to what I would call a decision infrastructure architecture.”
The closest historical parallel is the rise of search engine optimization in the early 2000s. Merchants who structured their sites for Google built compounding visibility advantages that lasted years. Those who ignored SEO became functionally invisible to an entire generation of online shoppers. The same dynamic is forming now with AI agents and Generative Engine Optimization (GEO), but the optimization target has changed. Instead of keywords and backlinks, agents evaluate structured product attributes, fulfillment rules, and trust signals. Merchants who invest in product data enrichment now will compound that advantage as agent adoption accelerates.
The feed is no longer a syndication artifact. It is your agent-facing operating system.
Suhas Gudihal
CTO of Athos Commerce
What agents actually need from your catalog
AI agents assess products across six categories of data. Most product catalogs deliver on one or two of them at best.
“Agents consume structured attributes. They consume availability. They consume policies. They consume trust signals. They consume fulfillment constraints. They consume relationship metadata,” Gudihal says.
Consider a shopper who asks an agent to find a lightweight carry-on bag under $150 with a laptop sleeve. To match that query, the agent needs structured attributes (weight, dimensions, compartments), availability (in stock, ships to the shopper’s location), policies (return window, warranty), and trust signals (ratings, review count). A catalog with only a product title, price, and marketing description can’t answer any of those questions with confidence, and the agent moves on.
This is where the distinction between agent-ready data and human-facing content becomes critical. Agents evaluate normalized attributes, deterministic fulfillment rules, and canonical product identifiers, while human shoppers respond to brand positioning, emotional resonance, and visual merchandising. Trying to serve both audiences from the same data layer degrades both experiences, which is why Gudihal argues, “You need to design for verifiability, not persuasion.”
The 30–44 age demographic already reflects this dual expectation. In the Athos Commerce/Pixel survey, 44.7% of this group said they want personalized product recommendations, and 37% want smart search with natural language processing.1 These shoppers are comfortable with AI-assisted discovery and traditional browsing. Merchants who build for both will capture this audience, and the window to get ahead of that expectation is still open.
The compounding cost of waiting
Product data quality compounds. Enrichment done today improves on-site search relevance, off-site feed performance, and agent discoverability at the same time. Every week of delay is a week your competitors build an advantage that becomes harder to close.
A practical place to start is auditing your top 50 products. How many structured attributes does each one have? Can an agent answer basic shopper questions about size, compatibility, return policy, and availability using only your feed? Most merchants who run this exercise for the first time find significant gaps, and those gaps represent revenue that’s already going to competitors with richer catalog data.
Platforms that unify on-site discovery, feed management, and catalog enrichment in a single system can accelerate this work, turning months of manual effort into weeks of automated improvement.
If an AI agent searched your catalog tonight on behalf of a shopper, would it find what it needs to recommend your products? Or would it recommend your competitor’s instead?
FAQs
What is agentic commerce?
Agentic commerce refers to the use of AI-powered shopping agents that research, compare, and purchase products on behalf of consumers. Unlike traditional ecommerce, where shoppers browse websites and make their own decisions, agentic commerce delegates part or all of the shopping process to an AI agent. Examples include Amazon Rufus, OpenAI Operator, Perplexity Instant Buy, and tools built on Google’s Universal Commerce Protocol. By Cyber Week 2025, AI agents influenced 20% of all global orders.
What is the difference between GEO and AEO?
Generative Engine Optimization (GEO) focuses on making your content and product data visible to generative AI tools like ChatGPT, Gemini, and Perplexity. Answer Engine Optimization (AEO) is a related discipline that focuses on structuring content so AI systems can extract direct answers to user questions. Both disciplines share a common goal: ensuring your products and content appear when AI systems respond to shopper queries, rather than only when humans visit your website.
What kind of product data do AI agents need?
AI agents evaluate products across six categories: structured attributes (size, weight, materials, features), availability (stock status, shipping regions), policies (returns, warranties), trust signals (ratings, review counts), fulfillment constraints (delivery timelines, pickup options), and relationship metadata (compatible products, bundles). Most product catalogs today provide only a fraction of these data points.
How is agentic commerce different from traditional ecommerce SEO?
In traditional SEO, merchants optimize for keywords, backlinks, and page structure to rank in search engine results. In agentic commerce, the optimization target shifts to structured product data. AI agents don’t read marketing copy or evaluate page design. They assess machine-readable attributes, fulfillment rules, and trust signals to determine whether a product matches a shopper’s query. Merchants who enrich their catalogs with structured data now will build compounding visibility advantages as agent adoption grows.
What is the difference between traditional ecommerce and agentic commerce?
In traditional ecommerce, the merchant controls the shopping experience through website design, search results, and merchandising. The website is the storefront. In agentic commerce, AI agents control the shopping experience on behalf of consumers, and the merchant controls the data. Visibility depends on structured product attributes, fulfillment rules, and trust signals rather than page design or paid media. The optimization discipline shifts from SEO (keywords and backlinks) to GEO and AEO (structured data and direct answerability).
How do I make my product catalog AI-ready?
Start by auditing your top 50 products. For each one, check whether an AI agent could answer basic shopper questions about size, compatibility, return policy, and availability using only your product feed. Then enrich each product across the six categories agents evaluate: structured attributes, availability, policies, trust signals, fulfillment constraints, and relationship metadata. Prioritize your highest-traffic and highest-revenue products first, and expand product data enrichment across the catalog over time.
Where should merchants start with catalog enrichment for AI agents?
A practical first step is auditing your top 50 products. For each one, check whether an AI agent could answer basic shopper questions about size, compatibility, return policy, and availability using only your product feed. Identify the gaps between what your catalog currently provides and the six data categories agents evaluate. Prioritize enriching your highest-traffic and highest-revenue products first, then expand across the catalog over time.
Sources & Further Reading
- Athos Commerce and The Pixel. “The Discovery Gap: What 800 Shoppers Reveal About Product Discovery.” Research report, January 2026.
- SparkToro/Datos. “State of Search Q1 2025: Behaviors, Trends, and Clicks Across the US & Europe.” Datos, 2025. https://datos.live/report/state-of-search-q1-2025/.
- Gartner. “Gartner Predicts Search Engine Volume Will Drop 25% by 2026, Due to AI Chatbots and Other Virtual Agents.” Gartner Newsroom, February 19, 2024. https://www.gartner.com/en/newsroom/press-releases/2024-02-19-gartner-predicts-search-engine-volume-will-drop-25-percent-by-2026-due-to-ai-chatbots-and-other-virtual-agents.
- Perez, Sarah. “OpenAI Launches Operator, an AI Agent That Performs Tasks Autonomously.” TechCrunch, January 23, 2025. https://techcrunch.com/2025/01/23/openai-launches-operator-an-ai-agent-that-performs-tasks-autonomously/.
- Kif Leswing. “Perplexity Announces Free Product to Streamline Online Shopping.” CNBC, November 19, 2025. https://www.cnbc.com/2025/11/19/perplexity-ai-online-shopping-paypal.html.
Find Which Athos Commerce Product is Right for You
