Rufus is quietly showing your Sponsored Products for free - and why your listings still lose scanners
Amazon has been placing Sponsored Products inside Rufus conversations, and most advertisers do not know it is happening. The feature is called Sponsored Products Prompts: when a shopper asks Rufus something like "Does this brand have a plant-based dish soap?", your ad can show up as the conversational answer - not a banner, not a search result, but a response inside an AI chat.
Amazon's AI auto-generates two kinds of prompts from your detail pages, Brand Store, and campaign data: comparison-style ("Why choose [brand] [category]?") and feature-specific ("Does [brand] have a [product] with [feature]?"). Campaigns are enrolled automatically. One seller's report for January 5 to February 14, 2026 showed 35 prompt entries across 18 unique questions, roughly 5,000 impressions, 41 clicks, $0.00 in spend (it appears free during beta), and 4 orders totaling $512.71 in sales. That is $512 in sales at zero ad spend, in a beta most advertisers have not noticed.
The catch: the Prompts tab inside an ad group usually shows "no data available" - you have to download the SP Prompts Report to see the real data. Actions to take now: pull your report, study the questions Amazon generates for your products, toggle off irrelevant prompts, and make sure your detail page reinforces the prompts that are spot on. The prompts also reveal how Amazon's AI understands your product - free competitive intelligence. One prediction worth heeding: right now Prompts is just an on/off switch with no bidding or targeting, but Amazon will eventually turn it into an auction, and brands that studied their data during the free beta will have a head start.
Separately, a reminder about why listings and A+ Content lose shoppers: most sellers build listings like product manuals, but the customer is mid-scroll, half-distracted, one thumb away from a competitor. Shoppers do not reject value; they reject effort. Nobody reads your listing - they scan it in an F- or Z-pattern, hunting for a reason to stay or bounce. The eye is drawn to contrast (big vs. small, bold vs. light) and avoids three things: too much text, too little structure, and no obvious point. Annoyance is a faster emotion than curiosity - you will lose them to frustration before you win them with information.
The fix is two-layered. Design for the scanners first: white space, font hierarchy, and visual contrast so key points pop instantly. Then write for the readers who slowed down: clarity and one clear point per section. Make headline and supporting text do different jobs - the headline grabs attention, the subtext builds buying confidence; if both say the same thing, you are wasting prime real estate. Every content element should follow a three-step sequence: scan (earn attention), read (reward with clarity), then buy (give them something structured enough to act on).
A useful framework for anyone working with affiliates and creators: stop organizing the landscape by tool type and follow the creator journey instead, across five layers. Layer 1, creator commerce intelligence - where creators decide what to promote (discovery tools, seller networks, seeding platforms); if you are invisible here, creators do not know you exist. Layer 2, demand creation - where creators influence shoppers before Amazon ever sees them; this is the external traffic Amazon keeps rewarding. Layer 3, conversion infrastructure - the links, widgets, and CTAs that turn influence into trackable clicks; broken pipes mean lost attribution. Layer 4, tracking and monetization - how traffic gets credited and everyone gets paid; this is the difference between a profitable program and lighting money on fire. Layer 5, performance optimization - the feedback loop that tells you what to scale or kill. Brands winning with creators operate across all five layers, not just "do affiliates."
Tool note: Google Labs has upgraded its free AI marketing tool Pomelli with a feature called Photoshoot - upload even a rough phone snap of a product, pick a template, and it generates professional-grade images in your brand style, powered by Google's image model. For sellers, the obvious uses are weak secondary image slots, fresh ad creative, and testing launch concepts before paying for a real shoot. It is US- and Canada-only for now, and free. It will not replace a photographer, but it fills content gaps at zero cost.
Quick hits: Walmart says roughly half its app users have tried Sparky, its AI assistant, and Sparky users build product baskets 35 percent larger than normal; TikTok Shop's U.S. GMV grew 68 percent in 2025 to $15.1 billion; direct mail, often declared dead, is still returning about $12 for every $1 spent; and one industry estimate puts total global e-commerce on a path toward $10 trillion.
