Shelf Life | Vol. 38 – The Invisible Heist: How AI Agents Are Stealing Your Customer Relationship
A consumer opened ChatGPT last week and typed: "Find me the best running shoes under $150, good for wide feet, available for delivery by Thursday."
ChatGPT returned three options with a ranked comparison. The consumer picked one. Done. No Google search. No brand website visit. No scroll through your carefully A/B tested PDP. No retargeting ad. No abandoned cart email sequence. The entire purchase journey happened inside a conversation with an AI, and your brand either showed up in that answer or it didn't.
That's not a hypothetical. That's Tuesday.
McKinsey estimates AI agents could mediate $3 to $5 trillion in global commerce by 2030. Adobe found that AI-driven e-commerce traffic surged 693% during the 2025 holiday season alone. ChatGPT now handles roughly 50 million shopping queries every single day.
The customer relationship you spent years and millions building? An LLM just put itself in the middle of it. The brands that understand what that means right now will own the next decade of retail. The ones still perfecting their homepage hero image will wonder where their conversion rates went.
Top Shelf Insights
🤖 AI agents could mediate $3 to $5 trillion in global commerce by 2030, according to McKinsey. Shopping through ChatGPT, Claude, Gemini, and Perplexity is already happening at scale.
🔍 Gartner predicts traditional search engine volume will drop 25% by 2026 due to AI chatbots and virtual agents. SEO built your business. AEO (Answer Engine Optimization) will determine whether you stay in it.
📦 Brands lose control of discovery and consideration when AI agents shop on a customer's behalf. Fulfillment, post-purchase experience, and supply chain are the surfaces you still own. Use them.
🏁 First-mover advantage is closing fast. Brands training AI on their product data today will be the brands AI recommends tomorrow. This is not a 2027 problem.
⚙️ According to Gartner, 60% of brands will use agentic AI to deliver personalized interactions by 2028. The brands that wait for the playbook will be handing market share to the ones who wrote it.
The New Middleman: How The Funnel Got Flipped
The traditional purchase funnel was built on one assumption: the customer comes to you. They search, they browse, they consider, they convert. You engineered every touchpoint. You controlled the narrative, the imagery, the reviews they saw first, the cross-sell, the urgency trigger. It was your stage.
In an agentic commerce world, the customer delegates. They open ChatGPT, Claude, Gemini, or Perplexity and describe what they need in plain language. The LLM surfaces products based on structured data, authority signals, and machine-readable information. It assembles a shortlist, applies the customer's stated preferences and budget, and either recommends or transacts. The customer confirms. The box ships.
Traffic to US retail sites from GenAI browsers and chat services increased 4,700% year-over-year in July 2025, according to Adobe. That number sounds exciting until you realize most of that traffic never reaches your site at all. The LLM handled the discovery. Your website was irrelevant to the decision.
The agent doesn't care about your hero image. It cares about your data.
Here's what changed at every stage of the funnel:
🔎 Discovery used to mean Google rankings. Now it means whether your product data is structured well enough for ChatGPT or Gemini to parse, understand, and surface in a relevant conversation.
🤔 Consideration used to mean your PDP, reviews, and retargeting ads. Now it means how cleanly your attributes, comparisons, and differentiators are formatted for machine reading inside an LLM response.
🛒 Purchase used to happen on your site, in your checkout flow, with your upsells. Now it happens inside the chat. OpenAI has checkout integrations with Target, Instacart, and DoorDash. Microsoft's Copilot Checkout turns conversations into conversions without a single redirect.
📦 Post-purchase is where you still own something. More on that in a minute.
The Data Problem: What LLMs Prioritize Versus What You've Been Building
Here's the uncomfortable reality for every brand that spent the last decade optimizing for Google. LLMs don't rank pages. They synthesize structured data, authority signals, and machine-readable product information. The entire discipline of SEO, built on keywords, backlinks, and click-through rates, transfers only partially to the world of AI-mediated commerce.
Gartner's Strategic Predictions for 2026 state that "traditional SEO and pay-per-click will give way to agent engine optimization" and that "products will need to be machine-readable." This shift is what AEO, Answer Engine Optimization, actually means in practice.
What Google cared about: 🔗 Backlinks and domain authority 📝 Keyword density and metadata ⏱️ Page speed and mobile optimization 🖼️ Rich content and engagement signals
What ChatGPT, Claude, and Gemini care about: 📊 Structured, clean, machine-readable product data 🧩 Entity relationships: how your products connect to categories, use cases, and customer needs ✅ Accuracy, consistency, and freshness of information across every channel 🗣️ Natural language clarity: can the LLM describe your product accurately without guessing? ⭐ Trustworthiness signals: reviews, third-party citations, factual authority
As CIO Magazine noted: "Clean catalogs, consistent metadata, and real-time inventory feeds will determine if you are even visible to AI shoppers. Brands will need to embed their differentiation into the data itself."
The brands winning in agentic commerce right now are the ones whose product catalogs read like a well-organized database. That is a fundamentally different content strategy than most teams have ever built, and it requires buy-in from every part of the organization that touches product data.
The Control Paradox: Where You Lost It And Where You Still Have It
When an AI agent shops on your customer's behalf, you lose control of five things you used to own:
🚫 The brand story and emotional narrative 🚫 The sequence of information the customer receives 🚫 The comparison set they're evaluated against 🚫 The moment and context of the purchase decision 🚫 The upsell and cross-sell during the shopping journey
But here's the paradox: the heist isn't total. You still own the back half.
McKinsey notes that at advanced levels of agentic commerce, "competition shifts from winning a single purchase to earning a place in the agent's ongoing plan," and that merchants need "deeper integration around loyalty, eligibility, substitutions, and service guarantees."
The agent selected you. Now you have to earn the customer.
📦 Fulfillment is your new first impression. When the agent places the order, the next human touchpoint is your packaging, your delivery speed, your unboxing experience. Supply chain reliability is no longer a back-office metric. It's a brand statement.
🔄 Returns and service define loyalty. The agent may have bypassed your marketing entirely, but a seamless return or a proactive customer service interaction is now how you get back into the consideration set.
📧 Post-purchase is the new acquisition channel. You earned a customer. Now you have to give them a reason to authorize their agent to come back to you. The email that lands after delivery, the loyalty program that feeds data back into the agent's preference layer, these are now your primary marketing surfaces.
The Playbook: Three Principles For Getting Agentic Commerce Right Now
Gartner research notes that retailers are "woefully unprepared for the shift to autonomous agentic AI" and warns that "many retail data sets are not yet modernized." The window to build ahead of this is open now. It will not stay open.
Principle One: Get Your Data House In Order (Yesterday)
The problem now is an organizational discipline one.
AI agents can only recommend what they can understand. If your product data is inconsistent across channels, if your attributes are incomplete, if your inventory isn't real-time, the agent will recommend your competitor whose data is cleaner. It's that simple and that brutal.
What this means operationally: 🗂️ Audit your product catalog for completeness, consistency, and machine-readability across every channel 🔄 Standardize attributes so that size, material, use case, and compatibility are expressed identically everywhere ⚡ Connect real-time inventory so agents don't recommend products you can't fulfill (a death sentence for repeat selection) 🧠 Enrich descriptions in natural language that answers real customer questions, not marketing language that sounds good to humans but means nothing to an LLM
According to commercetools: "Structured data, enriched metadata, and clean catalogs determine whether an agent can understand and recommend a SKU."
This is infrastructure work. It is not glamorous.
Principle Two: Shift From SEO To AEO Across The Entire Organization
This is not a marketing department initiative. This is a business model shift that needs executive ownership and cross-functional execution.
Gartner's Senior Principal Researcher Emily Weiss stated: "This marks the end of channel-based marketing as we know it." Gartner's research further notes that "marketers must prepare by putting strong data governance in place, tracking customer journey changes weekly, and integrating agentic systems into martech stacks."
Answer Engine Optimization means building your entire content and data strategy around how AI agents interpret, surface, and recommend your products inside LLM conversations. The practical translation:
👥 Every team that creates content or data touches your AEO strategy. Merchandising, marketing, e-commerce, supply chain, and IT all have a role. Most of them don't know it yet.
📐 Optimize for the question, not the keyword. AEO content is built around how customers describe real-world needs inside a chat conversation. "Lightweight waterproof jacket with hood under $150 that ships in two days" is more valuable to an LLM than any keyword you've been targeting.
🏷️ Build a knowledge base your agents can cite. Shopify's Agentic Storefronts let brands define their schema, FAQs, and brand voice so that when a customer asks ChatGPT about your product category, the LLM has accurate, brand-controlled information to draw from. That's the new homepage.
📊 Measure AI visibility, not just web traffic. If you're not tracking how your brand appears in ChatGPT, Gemini, Perplexity, Claude, and Copilot responses, you are flying blind in the channel that is growing fastest.
Principle Three: Train AI For Your Brand Before Someone Else Does
Gartner urges CMOs to "create unique branded research so their brands surface in GenAI results and AI overviews." This is the first-mover principle, and it matters more than most executives realize.
LLMs learn from what exists. The brands that are well-represented in authoritative content, structured product feeds, third-party citations, and AI-ready data today are the brands LLMs will default to recommending tomorrow. The brands that wait will be training AI on a landscape that already favors their competitors.
What first-mover advantage looks like in practice: 🎯 Feed your brand narrative into every machine-readable channel. Press coverage, structured product feeds, authoritative review platforms, and partnerships with AI commerce platforms all train LLMs on who you are. 🤝 Integrate with the platforms now. Walmart is already live with Google's Gemini. Shopify merchants are auto-enrolled in Copilot Checkout. OpenAI has checkout partnerships with Target, Instacart, and DoorDash. If you're not in these ecosystems, you're not in the consideration set. 🔁 Build your own agent experience. Brands like Ralph Lauren and Estée Lauder are already deploying their own AI shopping assistants. A brand-owned agent that knows your products, policies, and customer base is a direct counter to generic shopping agents making decisions about your brand without your input.
Gartner predicts 40% of enterprise applications will include task-specific AI agents by the end of 2026, up from less than 5% in 2025. The infrastructure for agentic commerce is being built right now by platforms you already use.
On the House
The thing that keeps me up about agentic commerce isn't the technology. The technology is moving fast but it's understandable. What's harder is the organizational implication.
Most retail and CPG companies are not structured for this shift. Their data teams sit in IT. Their content teams sit in marketing. Their supply chain teams sit in operations. Nobody owns the intersection of "is our product data machine-readable enough to win in an AI channel." That gap is where market share goes to disappear.
The brands that will win in agentic commerce are the ones that recognize this is not a marketing initiative or a tech project. It's a fundamental reassessment of how the customer journey works and who owns each part of it. When the discovery and consideration phases move inside an AI, you need a new answer to the question: where does our brand actually touch the customer?
The answer right now is: in the data before the purchase, and in the experience after it. Everything in between belongs to the agent.
That's not a crisis. It's a strategic reset. The brands that treat it like one will find themselves with a leaner, more efficient acquisition model and a post-purchase experience that actually drives loyalty. The brands that keep optimizing for a funnel that no longer exists will spend the next three years wondering where their conversion rates went.
The Last Look
If AI agents make purchase decisions based on data quality and machine-readable attributes rather than brand equity and emotional storytelling, does that fundamentally commoditize every category where product specs are comparable? And if so, where does brand differentiation actually live in an agentic world?
More to come in the Shelf Life series.
Follow me here for sharp takes on the trends shaping retail, fashion, and consumer product companies.
Want to talk more about how Gartner Consulting can help your organization?
Follow me on LinkedIn, Substack, or @ShelfLifebyJKS on Instagram or reach out!
📍 Jackie Swanson is a Managing Partner at Gartner Consulting, specializing in retail, consumer products, and utilities. She advises companies on large-scale transformations spanning strategy, operations, and technology. Jackie lives in New York with her husband and their three children.
#AgenticCommerce #AIRetail #AnswerEngineOptimization #BrandStrategy #RetailTransformation #DigitalCommerce #GartnerConsulting #ShelfLife