TL;DR
Nearly a third of U.S. shoppers now use AI tools to inform purchase decisions, and over half of consumers say they’ve replaced traditional search with generative AI for at least some product research. Brands that don’t appear in AI-generated answers are increasingly invisible at the exact moment people are deciding what to buy this post breaks down the numbers and the practical steps brands can take in response.
For three decades, “product discovery” meant one thing: type a query into a search box, scroll a page of blue links, click through a few, compare. That model is breaking down in 2026. A growing share of shoppers now describe what they want in a conversation with an AI system ChatGPT, Gemini, Perplexity, an AI Overview and get back a synthesized answer instead of a list of links to evaluate themselves. For brands, that’s a genuinely new discovery layer, and one most aren’t yet built for.
The numbers: how far the shift has actually gone
The scale of this shift is easy to understate if you’re not looking at the data directly.
- 30% of all U.S. consumers have used AI tools to inform a purchase decision, according to MarTech’s 2026 analysis of consumer research. Adoption skews younger but isn’t limited to Gen Z the breakdown is 40% of Gen Z, 42% of millennials, 28% of Gen X, and 13% of baby boomers.
- 43% of U.S. online shoppers used an AI assistant for product research in the past 90 days, and among them, 20% used AI on their most recent purchase over $50 meaning this isn’t just casual browsing, it’s shaping real spending decisions.
- 58% of consumers say they’ve replaced traditional search engines with generative AI tools for at least some product discovery, per consumer behavior research compiled by Karooli AI, drawing on Pew, Accenture, Deloitte, and other primary sources.
- 71% of Gen Z consumers use chatbots specifically for product discovery in e-commerce, according to Elfsight’s generational research the highest adoption of any cohort.
- 64% of consumers say they plan to use AI chatbots for shopping in 2026, according to a December 2025 survey of over 1,000 consumers by PartnerCentric — with 26% planning to use it more than they did the previous year.
- Shopping-related generative AI use grew 35% between February and November 2025 alone, per Boston Consulting Group research — one of the fastest-growing use cases for AI tools generally.
This isn’t a niche behavior anymore. It’s a meaningful, fast-growing slice of how real purchase decisions get made and it’s compounding, not plateauing.
Why people are switching — and what it means for trust
The “why” matters as much as the “how many,” because it tells you what brands actually need to supply.
Consumers say they turn to AI for shopping primarily to find the best price (55%), out of curiosity (46%), and for product discovery specifically (39%) with roughly a third also citing gift inspiration, per PartnerCentric’s survey. More broadly, research on AI and purchase journeys found that 45% of consumers now turn to AI at some stage of buying: 41% to research products, 33% to interpret reviews, and 31% to hunt for deals.
Critically, this isn’t blind trust. Only 54% of consumers say AI makes shopping easier, and while just 12% report regretting an AI-assisted purchase, a similar share say AI led them to spend more than they’d planned a double-edged effect worth brands’ attention on both sides. AI-referred traffic also appears to convert differently than traditional search traffic: HubSpot’s 2026 State of Marketing found that 58% of marketers describe AI referral traffic as high intent, even as 49% report their organic search traffic has declined because of AI-generated answers siphoning off clicks that used to land on their site.
That last stat is the uncomfortable part for a lot of brands: traffic isn’t just moving to a new channel, some of it is disappearing from the open web entirely, because the AI answer satisfies the question before a click ever happens.
The real-world evidence: it’s already showing up in referral data
This isn’t theoretical. Retail platforms are already seeing measurable AI-driven traffic. ChatGPT accounted for 20% of Walmart’s referral clicks in August 2025, up from 15% the month before a jump reported by Similarweb and covered widely (Digiday, Forbes, eMarketer, Modern Retail). It’s worth being precise about what that means: referral clicks are a specific, smaller slice of Walmart’s traffic under 5% of total site visits, according to multiple outlets covering the same data so this works out to roughly 1% of Walmart’s overall traffic, not a fifth of it. The number that matters isn’t the current scale, though, it’s the trajectory: a jump from 15% to 20% of referral clicks in a single month, on a channel that barely existed a year earlier.
Zooming out, Juniper Research projects worldwide retail chatbot-driven spending will climb from $12 billion in 2023 to $72 billion by 2028 a six-fold increase in five years.
What brands should actually do about it
The practical response isn’t to abandon SEO it’s to extend it. Most of the emerging guidance on this, sometimes called Answer Engine Optimization (AEO) or Generative Engine Optimization (GEO), builds directly on existing SEO fundamentals rather than replacing them.
1. Make your product information unambiguous and consistent. AI systems synthesize answers from what’s crawlable and citable. If your pricing, specs, or differentiators are inconsistent across your own pages let alone across retailer listings and review sites AI systems are less likely to cite you confidently. Consistency is now a ranking-adjacent signal, not just a customer-experience nicety.
2. Answer the actual decision-making questions, not just keywords. Traditional SEO optimized for what people typed into a search box. AI-era discovery optimizes for the question behind the question “what’s the best budget option for X,” “how does this compare to Y,” “is this worth it for Z use case.” Content that directly resolves those comparisons earns stronger visibility in AI-generated answers, according to HubSpot’s AEO research.
3. Invest in structured data and technical clarity. Schema markup, clear entity signals (making sure AI systems can correctly identify who you are, what you sell, and how you relate to competitors), and clean technical SEO fundamentals all make it easier for AI systems to parse and cite your content accurately, per multiple 2026 AEO guides from Amsive and Coalition Technologies.
4. Don’t neglect off-site presence. AI systems don’t only pull from your own website — Reddit threads, review sites, YouTube, LinkedIn discussions, and industry forums all shape how a brand gets described in AI answers, per Gray Bay Marketing’s 2026 AEO trend research. A brand with zero third-party discussion is a harder brand for an AI system to independently verify and recommend.
5. Track a different metric than click-through rate. Classic analytics undercount what’s happening here, since a successful AI-mediated interaction might never generate a click at all. Emerging metrics like answer inclusion rate how often your content is actually pulled into an AI-generated answer are a better read on whether this channel is working for you, according to Birdeye’s AEO research.
6. Treat this as additive, not a replacement. Google’s own 2026 guidance is explicit that core SEO practices remain relevant for AI Overviews and AI Mode, with no special AI-only markup required. The brands doing this well aren’t abandoning SEO they’re layering AEO/GEO practices on top of the same foundation.
The bottom line
Only 22% of marketers currently track their brand’s visibility in LLM-generated answers, according to research cited by The Gutenberg even though 74% already use AI for part of their own workflow. That gap between AI adoption and AI visibility tracking is the opportunity: most brands haven’t started measuring, let alone optimizing for, this discovery layer. The ones that start now are building an advantage in a channel that’s growing fast and getting more expensive to catch up on with every quarter that passes.
Related Buzz: We also covered [Beyond SaaS: Every “As a Service” Model Worth Knowing in 2026]