For most of the last two decades, a brand's homepage was the front door. Shoppers arrived, browsed a category, compared a few options, and worked their way toward a product. The job of marketing was to get them through that door, and the job of the site was to take it from there.
That sequence is coming apart. Shoppers increasingly describe what they need to an AI assistant and let it do the comparison work, which means the research, the shortlist, and often the decision all happen somewhere a brand does not own. By the time someone reaches the site, the choosing is mostly over.
On a recent Shopify webinar, "The New Rules of Brand Discovery," Matthew Rivard, Director of Strategy and Product Acceleration at Google, and Lore Oxford, Head of Insights Product at Reddit, joined Shopify's Technical Chief of Staff Matt Cohn to work through what that leaves brands to do. Google sees the discovery layer, Reddit sees the trust layer, and Shopify sees the commerce layer, and the three views converge on the same answer, one that is very much within a brand's control.
The reassuring part is how little of it is new. Be findable, be accurate, be worth recommending. Those rules have governed retail for as long as retail has existed. The shift is that they are now enforced by a reader that takes everything at face value.
To get recommended, a brand has to be legible to the machine and credible to the humans who use it. Everything else is execution.
The decision now happens before the click
The clearest evidence is where AI shoppers land. More than half of AI-referred sessions on Shopify start directly on a product page, compared with about 20% for organic search. Shopify first saw that pattern when it looked at AI referral traffic, and it has held consistently since. A shopper who lands on a homepage is still deciding. A shopper who lands on a product page has already decided and is checking.
That shopper is also worth more. Shopify's Q2 2026 commerce data shows that once an AI-referred visitor reaches a product page, they convert about 80% better than an organic-search visitor, and the channel is growing quickly: AI-referred sessions to Shopify storefronts grew 197% year over year, with orders up roughly 3x. Organic search still refers more sessions than every tracked AI platform combined and grew 12% over the same period on a much larger base, so this is a new surface rather than a replacement for an old one.
"The AI-referred shopper isn't just different. They're more valuable. They arrive more decided, they buy at higher rates, and they spend more when they do."
Rivard watches the same behavior form earlier, before a shopper has a product in mind at all. Google's role, as he describes it, is not only to answer the question someone already has.
"Google Search is where demand is created, not just fulfilled," Rivard said. Research Google ran with Ipsos found 71% of shoppers who come to Google Search are open to trying new brands or products, and brainstorming-style queries in AI Mode are growing about 30% faster than queries overall.1 People are arriving with a problem rather than a product, which is exactly the moment a brand can be introduced or overlooked.
What both vantage points describe is a decision that has moved upstream of everything a brand controls. Which raises the practical question the rest of the session was about: when a shopper never sees your homepage, what actually gets you into the recommendation?
AI can only recommend what it can read
The first requirement is the one most brands already know how to meet, and Rivard was direct about how little of it is new.
"You're still optimizing for the search experience, thus we still call this SEO," he said. "A good SEO strategy is the foundation for visibility in generative AI search experiences." Google's own documentation says it just as plainly: SEO best practices still apply to AI Overviews and AI Mode, including making sure the page meets the technical requirements for Google Search, following Search policies, and focusing on the key best practices, such as creating helpful, reliable, people-first content. The work that made a brand findable in search is the work that makes it findable in an AI answer, which is the premise behind Shopify's generative engine optimization playbook.
What has changed is the scale at which listings are read. Google applies its Gemini models to a Shopping Graph of more than 50 billion product listings, and more than 2 billion of them are refreshed every hour.2 At that volume, a listing is parsed and compared against alternatives, and the complete, accurate ones are the ones that qualify for a recommendation.
"For AI-powered shopping experiences to confidently recommend a product, they require structured, real-time product truth."
What earns a recommendation is accuracy that survives the click. When the price, availability, and shipping terms a shopper sees are the same ones they find on the product page, that consistency works in the brand's favor. Rivard's four priorities follow from that: complete data with rich titles and attributes, visual depth that shows products in use, operational transparency on stock and shipping, and demonstrated brand authority.
This is where the platform does real work. When AI drew on structured Shopify Catalog data instead of scraped or third-party feeds, the shoppers it referred converted 2x better. Shopify Catalog handles the technical layer automatically, standardizing, structuring, and enriching product data so it is machine-readable across AI channels without a separate integration per platform, and Agentic Storefronts manages that distribution from the Shopify admin. What stays with the brand is completeness and accuracy, which is the part no platform can do on anyone's behalf.
Trust is built beyond your website
Being readable gets a brand considered. Being trusted is what gets it recommended, and that judgment is shaped by human experience.
Oxford's framing is that this is not a new rule at all. People have always trusted other people more than they trust marketing, which is why influencers and authenticity dominated the last decade of the internet. What changed is that machines now inherit that instinct from human conversations about their lived experiences. No brand can ultimately dictate what a model says about it, which makes understanding why models value human conversation the more useful place to start.
"LLMs are obviously trained on huge amounts of human data, so what they find trustworthy is actually what people find trustworthy," Oxford said. "People really trust people."
Reddit put numbers to it. In a global study of about 32,000 respondents across four markets and 13 verticals, honest experiences from everyday people came out twice as trusted as AI-generated summaries when people decide what to buy, ahead of professional critics and influencers.3
Something else is happening on top of that, and it is new. As AI-generated answers become normal, people have started checking them.
"About half of people say they often or always feel the need to check Reddit to verify an AI's recommendations before they make a decision," Oxford said. "That's more than YouTube, that's more than TikTok, and it's actually more than friends and family as well." Roughly 40% of conversations on the platform are now commercial in nature,4 which means the verification layer and the discovery layer increasingly sit in the same place.
Rivard, listening to that, said it matched Google's own data, and he reached it from the opposite direction: 81% of people who discover a product on social media turn to Google Search to validate it.5 Two platforms with different business models, describing the same consumer instinct to check before buying.
Google's automated systems are designed to use many different factors to rank great content. After identifying relevant content, Google's systems aim to prioritize those that seem most helpful. To do this, they identify a mix of factors that can help determine which content demonstrates aspects of experience, expertise, authoritativeness, and trustworthiness, or what Rivard calls E-E-A-T.
"While E-E-A-T is not a single ranking factor, our systems use a combination of signals to recognize content that demonstrates high E-E-A-T," Rivard said.
So what does a brand actually do about a conversation it does not own? Oxford's advice was to start by listening rather than posting. Find out how people talk about you right now, she said, "if they think and feel anything about you at all." The point is diagnosis: where the conversations are happening, whether they are credible and useful, and whether they match what you want to be known for.
Oxford likens this to a dinner party, where the goal is to be a good guest who pays attention and follows the rules of the room. She recommends most brands begin with sponsored formats, because communities tend to respond better to a brand that is clearly acting as a brand.
"Today's conversations are going to inform tomorrow's decisions."
AI readiness is commerce readiness
Put the two requirements together and the list is one most teams could have written years ago. Accurate product data. Honest policies. Real reviews. A brand story specific enough to repeat. These are the fundamentals that have always separated good retailers from average ones. What has changed is the reward for doing them well, because the machine reading them now decides who gets recommended.
Shopify's category-level data shows the payoff concentrates where buying takes the most thought. In spec-led categories, where shoppers weigh specifications, compatibility, and tradeoffs, AI-referred shoppers convert at roughly twice the rate of organic-search shoppers. Apparel overall converts about 1.6x better from AI, and its research-heavy subcategories pull further ahead: watches about 2.4x, necklaces about 2.3x. The structured data that gets a product recommended is the same data that answers "does this meet my criteria" convincingly.
The practical version of this work breaks into three parts: the facts, the proof, and the language.
The facts are completeness and structure: every decision-relevant detail exists in a defined field rather than buried in a paragraph, variants and availability current, and policies published where a machine can find them.
The proof is social: reviews treated as always-on infrastructure rather than a quarterly campaign, plus the press, forum threads, and community discussion that corroborate what you say about yourself.
The language is the part most brands skip, and it is the part that will differentiate them as agents get better at interpreting softer signals: a specific brand story and a codified vocabulary, so an AI has something distinctive to represent.
Get the full checklist: how to prepare product data for AI channels
The advantage here is not solely about AI. A brand that gets its product data and its reputation in order becomes easier to find, easier to trust, and easier to buy from on every surface at once, including the ones that already work.
Get read, and get trusted
There is a version of this work that takes a year and a steering committee. There is also a version that starts this week.
Open an AI assistant and ask it about one of your highest-revenue products the way a customer would. Check whether the price, availability, variants, and return policy that come back are actually right. Then search your own brand name in a community platform and read what comes up without arguing with it. Between those two exercises you will have a fairly honest map of what machines can read about you and what people already believe, which is the whole job in miniature.
Fix the data first, because it is the part you control outright, and the fastest way to get filtered out of a recommendation is a price or a shipping promise that does not hold up. Earn the rest.
Frequently asked questions
What is AI-driven brand discovery?
AI-driven brand discovery is when shoppers use AI systems, like AI-powered search, chat assistants, and answer engines, to research, compare, and choose products before visiting a brand's website. Instead of browsing a homepage and category pages, the shopper describes what they want and the AI system recommends specific products. On Shopify, more than half of AI-referred sessions start directly on a product page, compared with about 20% for organic search, which shows how much of the comparison work now happens inside the conversation.
Is optimizing for AI search different from SEO?
Not fundamentally. According to Google, generative AI features are anchored in the same core ranking and quality systems as traditional search. Standard SEO fundamentals, meaning crawlable pages, helpful content, and accurate structured product data, remain the true drivers of discovery.
Why do community conversations influence what AI says about my brand?
Large language models are trained on large volumes of human conversation, so they tend to treat as trustworthy what people treat as trustworthy. In a Reddit study of about 32,000 people, honest experiences from everyday users were twice as trusted as AI-generated summaries when deciding what to buy,3 and Reddit is among the most-cited sources surfaced across major AI platforms. Google evaluates the same question through its E-E-A-T framework, which weighs firsthand experience, expertise, authoritativeness, and trust across the open web. Both amount to the same thing: while no platform can ultimately promise AI visibility, what independent sources say about you can shape what AI says about you.
Are AI-referred shoppers actually worth pursuing?
The early data suggests yes. On Shopify in Q2 2026, AI-referred shoppers converted about 80% better than organic-search visitors once they reached a product page, and about 2x better in spec-led categories. AI-referred sessions grew 197% year over year. These are early signals rather than guaranteed results for every brand, and organic search still drives more traffic in absolute terms, but the direction has been consistent.
What can my team do in the next 90 days?
Start with the data on your highest-revenue products, since that is what AI systems read to compare and recommend you: complete and structured attributes, current variants and availability, and policies published in machine-readable form rather than buried in the footer. When AI drew on structured Shopify Catalog data instead of scraped feeds, the shoppers it referred converted two times better. Then find out what people already say about your brand in community spaces. Treat it as a learning agenda and run small experiments rather than waiting for the playbook to settle.
Does my storefront still matter if shoppers are buying through AI?
Yes, arguably more. As comparison moves upstream, shoppers arrive with stronger intent and fewer touchpoints, which puts more weight on the surfaces you own. Product detail page quality, checkout, and the customer experience that earns good reviews all become more important, because they convert high-intent traffic and generate the trust signals AI systems read off-domain.
Read more
- AI and Organic Search Are Doing Different Jobs: What Shopify's Data Shows
- AI-referred Shoppers Convert Better and Spend More: What Shopify's Early Data Shows
- 8 Tips to Prepare Your Product Data for AI Channels
- The GEO Playbook: How (& Why) to Optimize for AI Discovery
Sources
- Google, "AI Mode: US insights" (71% open to new brands; brainstorming queries growing ~30% faster).
- Think with Google, "Google Shopping AI Mode and virtual try-on update": "The Shopping Graph now has more than 50 billion product listings… every hour more than 2 billion of those product listings are refreshed on Google."
- Reddit Path to Purchase 2026 Survey, US, UK, DE, AU, n=32,354, A18-65 (Monthly Social Media, LLM, and E-commerce users), Attest Sample, May 2026. All platform rankings statistically significant at a 95% confidence level against Social Platforms (YouTube, Facebook, Instagram, TikTok, Snapchat, X, Pinterest, Discord, LinkedIn) and All Platforms = Social Platforms, LLMs (ChatGPT, Gemini, Claude, Microsoft Copilot) and Amazon. Public summary: Reddit for Business.
- Reddit Insights powered by Community Intelligence, Global, 2024 vs 2025; posts with CI score >0.8.
- Think with Google, "AI-powered holiday season playbook" (81% of people who discover a product on social media validate it on Google Search).


