Computers Are Smart Now: Cracking Amazon's Rufus & Cosmo AEO with Jon Tilley
Amazon says shoppers who get a Rufus recommendation are about 60% more likely to buy. Jon Tilley of ZonGuru breaks down how to move from keyword stuffing to AI Engine Optimization (AEO).
Amazon shoppers who receive Rufus product recommendations have a 60% higher propensity to buy.
The archaic A9 keyword-stuffing algorithm has been replaced by Cosmo, an intelligent LLM that connects semantic dots.
Sellers must shift to AEO (AI Engine Optimization) by answering categorical intent questions naturally within their listing copy and imagery.
JT
Jon Tilley · Founder, ZonGuru
Jon Tilley is the founder of ZonGuru, an operational platform for Amazon brands and agencies. He specializes in reverse-engineering Amazon's search algorithms to maximize listing discoverability and sales.
The mini-framework
The Dual AI Discoverability Checklist
All required, any order
Semantic Mapping (Cosmo)
Feed the Product Knowledge Graph by naturally detailing what your product is, who it is for, when it is used, and what it pairs well with.
Rufus Q&A Optimization
Identify the top research and decision-making questions for your category, then seamlessly answer them in your listing copy and images.
The Single-ASIN Test
Update the title, bullets, and A+ content of a single listing first, measure the lift in discoverability and conversions, and then roll it out catalog-wide.
How this builds back to the Amazon Formula
Amazon says shoppers who get a Rufus recommendation are about 60% more likely to buy. That quietly changes the question every seller should be asking, from “am I ranking?” to “is the AI recommending me?”
Amazon’s old A9 algorithm matched keywords. It is being reshaped by Cosmo, the AI layer behind Rufus, Amazon’s conversational assistant. Cosmo doesn’t count keyword matches, it reads intent. Jon Tilley, who built ZonGuru by reverse-engineering Amazon search, calls the move SEO to AEO: stop stuffing keywords, start answering the question a shopper is actually asking.
Rufus shoppers are about 60% more likely to buy.
Jon on why that number matters:
Why Rufus Users Convert at 60% Higher Rates · 0:58 clipJon Tilley on why a Rufus recommendation lifts purchase intent so sharply.
Getting recommended comes down to what Jon calls the Dual AI Discoverability Checklist, two checks you can run in any order. First, semantic mapping: state plainly what your product is, who it’s for, when it’s used, and what it pairs with, so Cosmo can place you in its product knowledge graph. Second, the Rufus Q&A: find the top research and decision questions in your category, then answer them directly in your copy and images. Miss those signals and Rufus skips you, even when your product is the right fit.
Tell Cosmo what your product is, who it’s for, and what it pairs with.
Jon on what Cosmo actually reads:
Mapping the Product Knowledge Graph · 0:41 clipWhat Cosmo needs: what your product is, who it's for, and what it pairs with.
The upside is that this rewards better copy, not worse. Instead of cramming your title with disconnected keywords, you can write naturally and let the listing breathe, weaving the answers into your title, bullets, back-end terms, and image overlays. Start with one listing, measure the lift in impressions and conversions over a month, then roll it out across the catalog.
This is the Impressions lever of the Amazon Formula. Getting found used to mean ranking for keywords. Now it means being the answer the AI hands back, and that is the kind of shift that quietly decides who gets found next year.