Rufus Sponsored Prompts went live March 25. Your listing is now being read by an AI that decides whether to recommend you — and billing has started.
On March 25, 2026, Amazon quietly flipped a switch that changes how every product on the platform gets discovered.
Sponsored Products Prompts and Sponsored Brands Prompts moved from open beta to general availability in the U.S. Billing started immediately. Every active SP and SB campaign was automatically enrolled.
If you're running Amazon ads in the U.S. right now, you're already paying for this. Whether you know it or not.
Here's what happened — and why listing optimization just became the most important investment in your Amazon business.
What Are Sponsored Prompts?
Instead of clicking a keyword-triggered ad in search results, shoppers are now seeing AI-generated questions alongside their search results and on product detail pages:
"Does this provide thermal protection?" "Which TV is best for gaming under $300?" "Is this safe for sensitive skin?"
When a shopper clicks one of these prompts, Amazon's AI assistant Rufus generates a conversational answer — and your product may appear inside that answer as a recommendation, with an add-to-cart button right there.
This isn't a banner ad. This isn't a search result. Your product is being recommended by Amazon's AI as the answer to a shopper's specific question.
That's a fundamentally different trust signal than appearing in a grid of sponsored results.
Why This Changes Everything
Here's the part most sellers haven't processed yet:
You don't write the prompts. Amazon's AI does.
The prompts are generated automatically from your product detail page content — your title, bullets, description, A+ Content, backend keywords, and even your reviews. Amazon's AI reads your listing, decides what questions your product can answer, and creates the prompts.
You can see which prompts exist for your campaigns. You can pause individual prompts. But you can't create custom ones. You can't bid on specific prompts. You can't target them.
The only lever you have is the quality of your listing content.
Let me say that again because it's the entire point:
The only way to influence whether Rufus recommends your product is to make your listing worth recommending.
The Numbers That Should Wake You Up
Rufus now serves over 250 million monthly active users across the Amazon app and website. Amazon reported that Rufus drove $12 billion in incremental annualized sales in 2025.
But here's the stat that matters most: Research from the Mars Agency found that only 22% of products ranking on page one of Amazon search also appeared in Rufus recommendations. And 36% of the products Rufus recommended weren't even on page one.
Read that again.
Your page-one ranking doesn't mean Rufus will recommend you. And products you've never considered as competitors — products that don't even rank on page one — might be getting recommended instead of you.
The discovery game just split in two. There's search rank. And there's AI recommendation. They're not the same thing anymore.
Keywords vs. Prompts: The Shift
For 20+ years, Amazon product discovery has worked the same way: shopper types keywords → algorithm matches keywords → products appear in ranked order → ads push you up.
Prompts work on completely different logic.
A keyword match checks: Does this listing contain the words the shopper typed?
A prompt match checks: Can this product actually answer the shopper's question?
That's a different bar entirely. And most listings fail it. Not because they don't contain the right keywords — but because they don't communicate clearly enough for an AI to extract the answer.
Here's a real example. A haircare brand had Rufus generate these prompts for their product:
"Does this product provide thermal protection?" "Does it eliminate frizz?"
Those prompts were generated because somewhere in the listing content, the AI found thermal protection and frizz-related claims. But if the listing had buried those claims inside vague marketing language — "revolutionary formula for gorgeous hair" — Rufus would never have generated those prompts at all.
The product would be invisible in the AI discovery layer. Still ranking in search. Still running ads. But invisible to the 250 million people using Rufus.
What This Means for Your Listing
I've spent the last several months building a tool called PerfectASIN — a Chrome extension that runs deep diagnostic audits on Amazon listings, scoring them across every conversion lever: title, bullets, description, hero image, pricing, and competitive positioning.
The tool was built around a specific thesis: listing quality is the highest-leverage investment in your Amazon business.
Rufus just proved that thesis in a way I didn't anticipate.
Every module in PerfectASIN evaluates something that now directly affects your AI discoverability:
Title Analysis
Is your title structured so an AI can extract product identity, key attributes, and differentiators? Or is it a keyword-stuffed string that confuses both humans and machines?
Bullet Analysis
Do your bullets communicate specific, parseable product attributes? When Rufus scans your listing looking for answers to "Does this product do X?" — can it find a clear yes?
Description & A+ Content
This is where Rufus mines for deeper contextual answers. Use-case scenarios. Compatibility details. Ingredient breakdowns. If your A+ Content is just lifestyle photos with no text, Rufus has nothing to work with.
Hero Image Analysis
Main images influence click-through rates, but they also appear in Rufus recommendation cards. A weak hero image means a shopper skips past your product even after Rufus recommends it.
Price Intelligence
Rufus considers price in its recommendations. If you're overpriced relative to competitors for what the AI perceives as similar functionality, you won't be recommended for price-sensitive prompts.
What We're Building Next
Rufus has accelerated our roadmap. Three new features are in development for PerfectASIN v4:
1. AI Readability Score
A new sub-score within each module that evaluates how well your listing communicates to an AI reader — not just a human one. The score flags whether key product attributes are machine-parseable or buried in marketing fluff.
If Rufus needs to answer "Does this provide thermal protection?" and your listing says "advanced heat-defense technology for salon-quality results" — you might score well with humans but fail with AI. This score catches that gap.
2. Prompt Match Analysis
This will show you which Rufus-style prompts your listing would likely match — and which it wouldn't. Essentially: "Based on your listing content, an AI would recommend you for these queries... and would NOT recommend you for these."
This is a premium feature and, as far as I know, nothing like it exists in the market. It's the equivalent of seeing which keywords you rank for — but for AI recommendations.
3. Rufus-Aware Audit Framing
This is the quickest win. We're updating the audit narrative across all modules to reference the AI discovery shift. Instead of just "your title is weak," the report will tell you: "Your title lacks the attribute clarity that AI shopping assistants like Rufus need to recommend your product."
Same scoring engine. Bigger context. More urgency. Because now there's a concrete, revenue-impacting reason to fix that D-grade title — and it's called Rufus.
But Wait — It's Bigger Than Amazon
While everyone is focused on Sponsored Prompts, Amazon also expanded its Shop Direct program. This is the part that changes the entire e-commerce landscape.
Shop Direct lets Amazon customers discover and buy products from retailers who don't even sell on Amazon. Over 400,000 merchants. More than 100 million products. Rufus can now recommend off-platform products, link shoppers directly to a merchant's website, and even complete the checkout through a "Buy for Me" feature where Amazon's AI handles the purchase on the third-party site using the shopper's stored Amazon payment info.
Amazon is positioning itself as the starting point for every online purchase — not just purchases on Amazon.
And the engine driving all of it is Rufus. Which reads product content to decide what to recommend.
The same content optimization principles that drive Rufus recommendations on-Amazon will increasingly apply to off-Amazon discovery too. If your product content isn't clear, specific, and AI-parseable — you're invisible on both sides.
What You Should Do This Week
-
Check your Prompts tab. In your Ads Console, navigate to Campaign → Ad Group → Ads → Prompts. See what prompts Amazon generated for your products. If you see nothing — your listing content isn't giving Rufus enough to work with. That's your diagnosis.
-
Read your listing like an AI. Open your PDP and ask yourself: if a machine had to answer "What does this product do?" and "Why is it better than alternatives?" using only the text on this page — could it? If the answer involves interpreting marketing language, you have a problem.
-
Audit your bullets for attribute clarity. Every bullet should contain at least one specific, extractable product attribute. Not "premium quality materials" — but "316L surgical-grade stainless steel." Rufus can work with specifics. It can't work with adjectives.
-
Fill every Seller Central attribute field. Including optional ones. Rufus uses structured product data, not just free-text listing content. Missing attributes mean missing signals.
-
Run a diagnostic. I built PerfectASIN specifically for this kind of analysis. The free tier gives you a full listing diagnostic across all six conversion levers. The $5,000 Full ASIN Audit report tells you exactly where your listing leaks — and in a Rufus world, listing leaks now mean AI invisibility.
The Bottom Line
For 20 years, Amazon was a search engine with a buy button. Keywords got you found. Ads pushed you up. The listing mattered, but you could compensate for a weak one with enough ad spend.
That era is ending.
Rufus doesn't show shoppers a grid of results and let them decide. It reads your listing, evaluates your product, and decides whether to recommend you. If your content isn't clear enough for an AI to understand — you don't exist in the fastest-growing discovery channel on the platform.
Keywords got you found. Prompts get you recommended.
The listing IS the ad now. And the brands that figure that out first will own the next decade of Amazon.
Dan Matejsek is the founder of RavingFans.ai and creator of PerfectASIN. 27 years of e-commerce experience. $572M in career online revenue. He currently consults with Amazon brands on listing optimization, advertising strategy, and AI-powered growth.
Consulting inquiries: ravingfans.ai · All articles: ravingfans.ai/blog
