AI shopping outperformed search results on Prime Day, shaking things up for Amazon.

      Amazon created Prime Day to encourage shoppers to stay within its marketplace. This year, AI shopping brought in new customers as a chatbot provided recommendations. According to Adobe Analytics, US consumers spent a record $26.4 billion across retail websites during the four-day Prime Day event. However, the more intriguing figure is not the total spend, but rather the source of the most valuable buyers. For the first time, customers who accessed retailers through an AI assistant were the most likely to make a purchase, as reported by GeekWire.

      Shoppers directed by AI chatbots were approximately 40 percent more likely to finalize a purchase compared to those arriving via search, email, or social media, noted Forbes. In contrast, just a year ago, this type of traffic was the least effective in terms of conversion. The turnaround in under a year is the significant highlight here.

      The data reflecting this change is substantial. Adobe monitors billions of visits to US retail websites, and its insights for Prime Day were notably skewed. On the first day, traffic from generative AI to retailers nearly doubled compared to the previous year, with an increase of 98.3 percent. These visitors acted more like shoppers rather than mere browsers.

      AI-referred shoppers spent 49.9 percent longer on websites and viewed 20.5 percent more pages. They added items to their carts at a 33 percent higher rate than those from traditional sources. Essentially, the channel that previously consumed a retailer's time is now its most effective.

      This trend extends beyond a single event. Data from Adobe indicates that AI-referred traffic to US retail websites surged by 393 percent year-on-year in the first quarter. By March, this channel was converting around 42 percent better than non-AI traffic, while AI visitors in early 2025 converted approximately 38 percent worse. The trend has shifted dramatically.

      The implications of a referral source for Amazon are significant. Amazon's model is built on being the starting point for product searches, where its advertising revenue thrives, making ads a key profit driver. However, the Adobe data suggests a shift in that starting point.

      If more shoppers begin their journey with ChatGPT, Gemini, or Perplexity and only engage with a retailer for checkout, the discovery phase elevates to the AI companies. Consequently, the associated revenues shift as well. The assistant, rather than the marketplace, becomes the primary interface.

      Amazon is not remaining idle; it is promoting its own Alexa+ assistant and Rufus shopping bot while reevaluating compensation for the AI models that power them. However, the Adobe data indicates that general-purpose chatbots are already fulfilling the initial discovery role, often before Amazon is involved.

      The traditional search paradigm is evolving. For two decades, users started their journeys on Google. The same forces influencing AI shopping are also challenging Google's dominance in search: people are increasingly turning to assistants instead of submitting typed queries. Marketers refer to this phenomenon as GEO, succinctly stating that AI search is the new SEO.

      For retailers, the key takeaway is practical. Product catalogs must be accessible to machines since chatbots can only recommend items they can process. Adobe notes that many retail websites remain poorly optimized for machine readability, leading to a disconnect between incoming traffic and the readiness of the stores to accommodate it.

      There is a broader competition for commerce as well, with OpenAI now positioning itself in the advertising sector. A chatbot capable of both recommending products and processing payments starts to resemble a marketplace.

      There’s an important distinction for international readers: the Adobe data pertains specifically to American retail, and Europe is lagging in this area. The adoption of shopping assistants is slower in Europe, and the EU's Digital Markets Act is already influencing how Google and Amazon rank and display products. If shopping assistants do emerge as the primary entry point in Europe, regulators may face unprecedented challenges, as the entities controlling the assistants would take over the role previously held by gatekeepers.

      This holds significant implications for European retailers and brands. The focus is shifting from a search interface they've learned to navigate to a system that's opaque. The AI laboratories are currently establishing the rules for this new market dynamic, mainly within the US and primarily on their own terms.

      However, two warnings should be noted. The data originates from a single vendor, Adobe, which sells both the analytics and the AI tools that benefit from this narrative. Independent data sources would provide a helpful counterbalance.

      The second warning pertains to scale. Despite the tripling of AI referrals, they still represent a minor portion of overall retail traffic. A channel may have high conversion rates yet still remain relatively small, and a four-day event does not reflect full-year performance. The trend is clear, but the scale is not yet substantial.

      Nonetheless, the landscape of ecommerce is quietly transforming. Previously, the focus was on how to rank on Google or secure the Amazon buy box. The new question is simpler yet more complex: when an assistant makes a recommendation, who owns the customer?