Amazon Optimisation
Unlocking Visibility: How to Thrive in Amazon’s AI-Powered Era
Amazon’s AI revolution is in full swing. With Rufus delivering conversational recommendations and COSMO powering intelligent search intent, alongside the evolving A10 system, sellers must adapt strategies to remain competitive.
28 July 2025
8 min read
1. Insights from Amalytix’s Rufus Study
Amalytix analysed what makes products "Rufus-recommended":
Titles: Nearly all top listings used the full 200-character allowance.
Images & videos: 6–9 images (median 7), at least 3 videos.
Bullet points: 5 clear and complete bullets performed best.
Enhanced content: ~55% used A+, ~32% used A+ Premium.
Ratings & reviews: Median rating was 4.5 with ~3,000 reviews.
Fulfilment: >94% FBA, >92% Prime eligible.
Badges: "Amazon’s Choice" was present on 42% of Rufus picks.
This data gives a clear picture of the kind of listings Amazon’s AI is trained to surface: detailed, media-rich, and trusted.
2. COSMO vs A10 vs Rufus: What Powers the Future of Amazon Search?
System: A9 / 10
Role: Classic search ranking
Focus: CTR, conversion, seller authority, off-Amazon traffic
System: COSMOS
Role: AI search engine
Focus: Semantic reasoning, knowledge graph integration, shopper intent
System: Rufus
Role: Conversational Layer
Focus: Natural language interaction + visual AI + COSMO-powered responses
What is COSMO?
COSMO (short for Contextual Shopping Model) is Amazon’s next-gen AI engine built to understand the why behind a search, not just the keywords typed.
Rather than matching “running shoes” to listings with exact terms, COSMO interprets broader intent:
Query context: It knows that “best shoes for flat feet” might not return the same items as “fastest shoes for 5k racing.”
Knowledge graph: COSMO integrates structured data about brands, product categories, uses, features, and even customer sentiment.
Semantic matching: It connects questions like “What’s a good sunscreen for sensitive skin?” to listings even if those exact words aren’t used, so long as the product actually fits the need.
In short, COSMO doesn’t just scan listings for words. It reasons like a shopper. This semantic understanding underpins both traditional search and new AI agents like Rufus.
And if your content isn’t aligned with this kind of reasoning, if it’s keyword-stuffed, vague, or lacking visual clarity, COSMO will bypass it entirely.
3. Four-Step Optimisation Playbook
Check your visibility with Rufus
Ask: “What’s the best product for…” your use case.
See if your listing appears. If not, you’re not in the AI’s frame.
Spy on competitor listings that are recommended
Review how they’ve structured their copy, content, images, and A+ assets.
Emulate structure, but be original in messaging.
Ask about your own product
In Rufus, phrase customer-style questions (“Does this protect against UVB?”) and assess what gets returned.
If your product isn’t clearly answering, your bullets or visuals likely need work.
Run your product and supplementary images through a piece of software like AWS Rekognition
This Amazon tool lets you check what AI can “see” in your product imagery.
Ensure your packaging, feature callouts, and product use are all recognisable, Rufus and COSMO both rely on visual cues.
4. Why It All Matters
COSMO is the engine of intent, your listing must make sense to it semantically.
Rufus is the interface, it pulls from COSMO’s logic to answer shopper questions.
A10 still matters, but is now just one layer of a much more sophisticated AI ranking system.
AI shopping agents will increasingly take the wheel in product discovery and if your content isn’t tailored to be interpretable by models like COSMO and Rufus, you’ll be invisible.
Final Word
The age of “optimise for keywords” is over. Now it’s about optimising for understanding.
To thrive:
Use natural language that matches real questions.
Make your visuals clear and AI-readable.
Create listings that answer why your product is the right choice.
You’re not just ranking for algorithms, you’re being judged by a reasoning engine.
LMO7 can help brands navigate this new world. From AI visibility audits to COSMO-optimised content creation, we’re building the playbook for the next generation of Amazon commerce.
Amalytix analysed what makes products "Rufus-recommended":
Titles: Nearly all top listings used the full 200-character allowance.
Images & videos: 6–9 images (median 7), at least 3 videos.
Bullet points: 5 clear and complete bullets performed best.
Enhanced content: ~55% used A+, ~32% used A+ Premium.
Ratings & reviews: Median rating was 4.5 with ~3,000 reviews.
Fulfilment: >94% FBA, >92% Prime eligible.
Badges: "Amazon’s Choice" was present on 42% of Rufus picks.
This data gives a clear picture of the kind of listings Amazon’s AI is trained to surface: detailed, media-rich, and trusted.
2. COSMO vs A10 vs Rufus: What Powers the Future of Amazon Search?
System: A9 / 10
Role: Classic search ranking
Focus: CTR, conversion, seller authority, off-Amazon traffic
System: COSMOS
Role: AI search engine
Focus: Semantic reasoning, knowledge graph integration, shopper intent
System: Rufus
Role: Conversational Layer
Focus: Natural language interaction + visual AI + COSMO-powered responses
What is COSMO?
COSMO (short for Contextual Shopping Model) is Amazon’s next-gen AI engine built to understand the why behind a search, not just the keywords typed.
Rather than matching “running shoes” to listings with exact terms, COSMO interprets broader intent:
Query context: It knows that “best shoes for flat feet” might not return the same items as “fastest shoes for 5k racing.”
Knowledge graph: COSMO integrates structured data about brands, product categories, uses, features, and even customer sentiment.
Semantic matching: It connects questions like “What’s a good sunscreen for sensitive skin?” to listings even if those exact words aren’t used, so long as the product actually fits the need.
In short, COSMO doesn’t just scan listings for words. It reasons like a shopper. This semantic understanding underpins both traditional search and new AI agents like Rufus.
And if your content isn’t aligned with this kind of reasoning, if it’s keyword-stuffed, vague, or lacking visual clarity, COSMO will bypass it entirely.
3. Four-Step Optimisation Playbook
Check your visibility with Rufus
Ask: “What’s the best product for…” your use case.
See if your listing appears. If not, you’re not in the AI’s frame.
Spy on competitor listings that are recommended
Review how they’ve structured their copy, content, images, and A+ assets.
Emulate structure, but be original in messaging.
Ask about your own product
In Rufus, phrase customer-style questions (“Does this protect against UVB?”) and assess what gets returned.
If your product isn’t clearly answering, your bullets or visuals likely need work.
Run your product and supplementary images through a piece of software like AWS Rekognition
This Amazon tool lets you check what AI can “see” in your product imagery.
Ensure your packaging, feature callouts, and product use are all recognisable, Rufus and COSMO both rely on visual cues.
4. Why It All Matters
COSMO is the engine of intent, your listing must make sense to it semantically.
Rufus is the interface, it pulls from COSMO’s logic to answer shopper questions.
A10 still matters, but is now just one layer of a much more sophisticated AI ranking system.
AI shopping agents will increasingly take the wheel in product discovery and if your content isn’t tailored to be interpretable by models like COSMO and Rufus, you’ll be invisible.
Final Word
The age of “optimise for keywords” is over. Now it’s about optimising for understanding.
To thrive:
Use natural language that matches real questions.
Make your visuals clear and AI-readable.
Create listings that answer why your product is the right choice.
You’re not just ranking for algorithms, you’re being judged by a reasoning engine.
LMO7 can help brands navigate this new world. From AI visibility audits to COSMO-optimised content creation, we’re building the playbook for the next generation of Amazon commerce.