AMAZON SEARCH & AI OPTIMIZATION
Keyword SEO Alone
No Longer Wins on Amazon
Rufus has 300 million users and generated $12 billion in incremental sales in 2025. COSMO powers how Amazon understands shopper intent. A10 still runs keyword search. Eva optimises for all three — as a connected system, not separately.
THE THREE-LAYER SYSTEM
Amazon Search Is Now Three Systems in One
Most Brands Optimise for Only the First
The way Amazon surfaces products has changed fundamentally. A10 still handles keyword-based search ranking. COSMO — Amazon’s commonsense intelligence layer — infers why shoppers want products and categorises listings based on intent, not just words. Rufus is the AI shopping assistant that reads your entire listing to answer natural-language questions and recommend products. Winning in 2026 requires all three working together.
THE PROBLEM
What Keyword-only SEO Misses in 2026
The way Amazon surfaces products has changed fundamentally. A10 still handles keyword-based search ranking. COSMO — Amazon’s commonsense intelligence layer — infers why shoppers want products and categorises listings based on intent, not just words. Rufus is the AI shopping assistant that reads your entire listing to answer natural-language questions and recommend products. Winning in 2026 requires all three working together.
WHAT EVA MANAGES
Five Layers of Amazon Search Optimisation
01.
Title & Noun Phrase Architecture
Amazon’s patent for Rufus explicitly highlights noun phrase optimisation as a core strategy. Eva writes titles that lead with the primary keyword (within the first 80 characters for mobile visibility), then build with descriptive noun phrases that give COSMO and Rufus rich context: product type, key material, use case, and distinguishing attribute. ‘Organic Green Tea — 100 Bags, Unsweetened, Japanese Sencha’ outperforms ‘Japanese Green Tea Bags Organic Matcha Tea Best Tea’ in every metric. Rufus parses semantic noun phrases. Keyword strings read as noise.
02.
Bullet Points — Intent-First, Question-Answering
Each bullet point should answer a probable shopper question in natural language. Not ‘Premium quality materials’ — but ‘The reinforced aluminium frame withstands drops up to 1.5 metres, tested for daily commuter use. Eva audits Q&A sections and one-star reviews to identify the top questions and concerns shoppers have, then ensures each bullet addresses at least two of those signals directly. This is the most critical Rufus optimisation lever and the most consistently underdone by standard SEO.
03.
Backend Attribute Completeness — COSMO Inputs
COSMO builds its categorisation of your listing from backend product attributes: subject, target audience, intended use, material, item type, and discovery fields specific to each product type. Most sellers leave these partially filled or blank. Eva completes every applicable backend attribute with precise, intent-aligned values — giving COSMO the data it needs to connect your product to the right shopper intentions. Data consistency is enforced across all fields: any conflict between title, description, and backend values triggers Rufus suppression.
04.
Q&A Seeding, Reviews & Multimodal Optimisation
Rufus reads Q&A sections directly to generate answers. Eva seeds
the Q&A with the top anticipated shopper questions — use case queries, comparison questions, and concern-based questions from competitor reviews — and provides detailed, conversational answers. Images are treated as data sources: alt text in A+ content is optimised for computer vision extraction, and vertical video (9:16 format) is produced for the Rufus mobile chat stream where video carousels surface in ecommendations. Review velocity and sentiment are monitored — Rufus synthesises reviews to inform its responses.
05.
Technical Health — Indexing, Variations & Suppression
Great content is invisible if Amazon cannot index it. Eva audits parent-child variation structures for broken parentage, identifies listing suppression across all ASINs, verifies backend keyword indexing, and resolves compliance flags. Shoppable A+ Content — with interactive carousels and in-content add-to-cart functionality —is implemented on top-revenue ASINs. Organic rank shifts are tracked continuously and content is updated as search behaviour evolves or Amazon introduces new compliance standards.
Rufus + Sponsored Ads
Sponsored Ads Now Appear Inside Rufus Conversations
In 2026, Sponsored Products and Sponsored Brands campaigns appear directly within Rufus conversations — reaching shoppers at the moment they are actively engaging with AI-driven product recommendations. This creates a direct connection between listing optimisation and advertising performance that most agencies manage as separate disciplines.
A listing optimised for Rufus — with clear noun phrases, intent-aligned bullets, and complete attribute data — performs better when it appears in Rufus ad placements than a keyword-stuffed listing does. Eva manages listing optimisation and Amazon PPC as one coordinated system: the same content improvements that make Rufus recommend your product also make your ads more effective when they appear in Rufus conversations.
THREE-LAYER VS. KEYWORD-ONLY
What Changes When All Three Systems Are Optimised
Capability
Primary optimisation target
Title strategy
Bullet points
Backend attributes
Data consistency
Q&A management
Image optimisation
Rufus ad performance
Technical audit
Eva Commerce
A10 + COSMO + Rufus
all three, coordinated
Noun phrase architecture —
semantic, readable, COSMO-clear
Intent-first answers to anticipated shopper questions
Complete — subject, target audience, intended use, all discovery fields
All fields reconciled — conflicts
suppressed by Rufus/COSMO flagged and fixed
Seeded with high-probability questions and detailed answers Rufus can cite
Alt text structured for computer vision; vertical video for Rufus mobile stream
Listing optimised to perform in Rufus ad placements as well
as organic
Continuous — suppression, parentage, indexing, compliance monitoring
Keyword-Only
Amazon managed service or reseller with markup and minimums
Keyword density , high-volume terms first
Feature lists with keyword repetition
Partially filled or blank
Not audited — conflicts common
Left empty or unmanaged
Designed for humans only
Not considered
Infrequent, reactive
Case Studies
Morinaga
Scaling Amazon Growth Across US and Japan with Eva’s Unified Commerce System
- Amazon
Cricket
The Cricket Company Doubles Amazon Ad Sales While Improving Profitability with Eva’s Commerce Growth System
- Amazon
Happy Products
Scaling Amazon Revenue and Profit with Eva’s Growth System
- Amazon
Maelys
Scaling TikTok Shop Revenue While Increasing Shopify Conversion and Profitability
- TikTok
- Shopify
Perupima
Scaling Shopify Revenue and Conversion with Eva’s Conversion Intelligence
- Shopify
View All Case Studies
Hear from the brands growing
Eva helped us align our Amazon operations across the US and Japan into one system. By connecting advertising, content, and inventory, we gained better visibility and control over performance.
This allowed us to improve conversion, stabilize operations, and scale more efficiently across both markets.
Teruhiro Kawabe
Chief Marketing Officer
Eva brought structure and intelligence to our Amazon advertising. By combining AI-driven optimization with expert execution, we were able to double our ad sales within six months while maintaining strong return on spend.
Tina Fichera
VP of eCommerce, The Cricket Company
I see Eva as the system that connects demand, ranking, and conversion across Amazon, Shopify, and TikTok. Their combination of AI and expert operators has significantly increased our visibility while improving profitability.
Juliette Fasset
CEO, Happy Products Inc.
“Eva helped us turn TikTok into a real revenue engine, not just a content channel.
What changed was the structure. Creator activity became measurable, scalable, and directly tied to revenue.
We moved from unpredictable spikes to consistent growth.”
Beth Milligan
Operations Director, Drybar
Eva enabled us to unify Amazon advertising, data, and execution into one system. This allowed us to scale demand efficiently while maintaining strong control over profitability across markets.
Linda Vargova
Head of eCommerce, MyProtein
Proven Results Across the Eva Growth System
Real outcomes from brands using Eva across Amazon, Shopify, and TikTok
Across Channels
with Profit Focus
in 6 Months
Operational Costs
in 6 Months
Based on aggregated performance across Eva-managed brands
Frequently Asked Questions
Does traditional keyword SEO still matter with Rufus and COSMO?
Yes — A10 still handles the majority of Amazon product searches. Keyword placement in your title, bullets, and backend fields remains essential for ranking in traditional search results. The shift is that optimising for keywords alone is now insufficient: COSMO penalises keyword stuffing, and Rufus requires semantic clarity and intent- aligned content to recommend your product. The three layers are complementary — optimising for all three outperforms optimising for any single one.
What is COSMO and how is it different from Rufus?
COSMO is Amazon’s backend commonsense knowledge framework — the intelligence layer that powers how Amazon understands shopper intent. It is not a ranking algorithm you can see; it builds connections between what shoppers search, what they actually want, and which products fit those intentions, using data from co-buy and search-buy behaviours. Rufus is the frontend AI shopping assistant that draws on COSMO’s understanding to answer natural-language questions and recommend products. Optimising for Rufus and optimising for COSMO are related but distinct tasks — COSMO is fed primarily by backend attributes and semantic content coherence; Rufus is fed by the entire listing including Q&A, reviews, and images.
Will Rufus optimisation hurt my keyword ranking?
No — it improves it. A10 heavily weights conversion rate. Listings that are more relevant and clearer — which Rufus optimisation produces — convert better . Higher conversion strengthens A10 ranking. The disciplines reinforce each other: semantic clarity improves Rufus recommendations, which drives higher-intent traffic to the listing, which converts at a higher rate, which strengthens A10 rank. Eva has consistently seen A10 ranking improve after Rufus and COSMO optimisation, not decline.
What is the 'Help Me Decide' feature and how does it affect listings?
Help Me Decide, launched by Amazon in October 2025, is a Rufus feature that recommends specific products with explanations drawn from listing data, customer reviews, and shopper history . It explicitly surfaces product rationale — why this product is the right match for the shopper’s stated needs. Listings with clear use-case statements, specific attribute data, and high-quality reviews are disproportionately featured. Q&A sections and A+ content are primary data sources for Help Me Decide recommendations.
300 million shoppers are asking Rufus what
to buy. Is your listing the answer?
Eva audits your current listings against all three systems — A10, COSMO, and Rufus — and shows you exactly where you are losing visibility that keyword reports cannot detect.
