How Amazon's A10 Algorithm Works (And What Brands Need to Do Differently in 2026)
Search “Amazon A10 algorithm” and you will find a thousand articles saying roughly the same thing: external traffic matters more, PPC matters less, organic sales are king. Almost none of them cite a primary source. Most recycle each other.
The reality in 2026 is more interesting and more useful than the version most agencies are selling. Amazon has never officially renamed its search algorithm from A9 to A10. What the seller community calls “A10” is shorthand for a series of observable shifts in how Amazon ranks products. Those shifts have compounded for years, and they accelerated once Amazon put AI in front of search: first with Rufus, its generative AI shopping assistant, and the COSMO knowledge graph, and as of May 2026, with Alexa for Shopping, the assistant that now combines Rufus and Alexa+ across the Amazon Shopping app and website.
If you are a brand operator trying to rank in 2026, you need to understand three things: what the algorithm actually optimizes for, what changed when AI search arrived, and which of the “A10 best practices” floating around online are now actively wrong.
This guide walks through all three.
What is the "A10" algorithm, really?
Amazon’s product search algorithm was originally developed by A9.com, a search subsidiary Amazon founded in 2003 and later folded into its retail organization. The name is a numeronym for “algorithm“: the letter A followed by nine more letters. Amazon has never publicly announced an “A10” successor. Articles claiming otherwise began circulating around 2020, and none of them cited a primary source.
What sellers and agencies call A10 is the cumulative effect of changes Amazon has made to its ranking system since roughly 2020:
- A reduction in the relative weight of pay-per-click (PPC) sales velocity
- Increased weight on organic sales, repeat purchases, and customer satisfaction signals
- Stronger credit for off-Amazon traffic that converts on-platform
- Tighter integration of seller account health into product visibility
- AI-driven query understanding (COSMO) and AI-mediated shopping (Rufus, now Alexa for Shopping)
Read it this way:
So when this article says “A10,” read it as “Amazon’s current search ranking system,” not as an Amazon-sanctioned product name. The distinction matters because most “A10 hacks” content treats the algorithm as a fixed object you can reverse-engineer once and exploit forever. It is a moving system, and the most important shifts of the last two years have nothing to do with A9 versus A10 framing at all.
A9 vs A10: what actually changed
The original A9 algorithm was famously conversion-obsessed. The simplified version sellers internalized was: get clicks, convert clicks into sales, let sales velocity drive ranking, and let ranking drive more clicks. PPC was the most reliable lever to spin that flywheel, and entire agencies were built around aggressive launch strategies that exploited it.
The shifts that earned the “A10” label changed that calculation in specific ways.
Organic sales count for more than PPC sales.
Sales attributed to organic search and external sources now contribute more ranking momentum than the same dollar volume attributed to a Sponsored Products click. This is widely observed across categories, though Amazon does not publish the ratio.
Click-through rate from search results matters more.
A listing that earns clicks at a higher rate than competitors in the same position tends to gain rank, even before those clicks convert. This rewards strong main images, compelling titles, and competitive pricing visible in the search grid.
Seller authority is a real input.
Account health metrics influence visibility: Order Defect Rate, Late Shipment Rate, Valid Tracking Rate, policy violations, and listing quality scores. A seller with deteriorating account health will see ranking decay even on listings with strong sales history.
External traffic that converts gets credit.
Traffic sent from Google, Meta, TikTok, email, or influencer content that lands on Amazon and converts contributes to ranking signals. This is the one piece of “A10 lore” that has held up over time, and Amazon puts money behind it: the Brand Referral Bonus program pays brand-registered sellers a bonus averaging 10% of qualifying sales driven from external channels, with the exact rate varying by category.
Repeat purchases and Subscribe & Save matter.
Products that build a base of repeat customers see compounding ranking benefits. Subscribe & Save subscriptions in particular carry weight for category-relevant consumables.
What did not change
If your title, bullets, backend search terms, and A+ Content do not tell Amazon what your product is, no amount of external traffic will rank you for queries you do not index for.
How the algorithm ranks products in 2026
Across categories and listing types, the factors below consistently move ranking. They are listed roughly in order of leverage, but the actual weighting is product- and category-dependent.
01
Relevance signals
Title, bullet points, description, A+ Content text, and backend search terms determine which queries you can rank for at all. In 2026, Amazon’s COSMO knowledge graph also infers relevance from implicit attributes — a query like “gift for a runner who just had knee surgery” can surface low-impact cross-trainers without those exact words appearing in the listing.
02
Conversion rate at the query level
03
Sales velocity, weighted by source
04
Click-through rate from search
05
Reviews and ratings
06
Pricing competitiveness
07
In-stock rate and Inventory Performance Index
08
Account health and listing quality
09
A+ Content, Brand Story, and Brand Registry status
10
External traffic that converts
How an Alexa for Shopping price alert flows across the Amazon app, phone notification and Echo Show. (Amazon Image)
Rufus, COSMO, and Alexa for Shopping
The most important development in Amazon search over the last two years is not an algorithm update in the traditional sense. It is the layer of AI that now sits between the shopper and the ranking system.
Rufus, Amazon’s generative AI shopping assistant, launched in beta in February 2024, reached all US customers on the app and desktop later that year, then expanded to international markets. Amazon says Rufus helped more than 300 million customers research, compare, and buy in 2025. In May 2026, Amazon folded Rufus and Alexa+ into a single assistant called Alexa for Shopping, available to all customers on the Amazon Shopping app and website. Shoppers can now ask conversational questions directly in the main Amazon search bar, and AI-generated overviews appear at the top of search results and on product detail pages.
COSMO is Amazon’s commonsense knowledge graph, detailed in a research paper Amazon published in 2024 and deployed in Amazon search. It maps the relationships between products, attributes, use cases, and shopper intent, and it is the reason Amazon search has become noticeably better at handling long, intent-rich queries that contain no product-category keywords.
Together, these systems change what “ranking” means. Your listing is no longer competing only for a position in a search grid. It is also competing to be the product an AI assistant selects, summarizes, and recommends. That has two practical implications.
First, keyword stuffing is a strictly worse strategy in 2026 than it was in 2022.
Amazon’s systems infer relevance from product attributes, structured data, and review content. A clean, accurate listing with rich attribute data and authentic review depth outperforms a keyword-bloated listing with shallow content, even on queries the bloated listing technically indexes for.
Second, review content is now part of your discoverability surface.
Reviews that mention specific use cases, problems solved, and contexts of use give COSMO and Alexa for Shopping the inputs they use to surface your product on conversational queries. That makes review acquisition more strategically important than it was under pure A9: Vine, Subscribe & Save reviews, and post-purchase flows that comply with Amazon’s communication policies.
What brands need to do differently
Most of the “A10 advice” still being repeated is years out of date. Here is what actually moves the needle now.
Stop optimizing titles for keyword density. Start optimizing them for click-through.
Build attribute completeness.
Treat reviews as a discoverability asset, not just a conversion asset.
Send external traffic to Brand Store URLs with Brand Referral Bonus tagging.
Defend account health like a ranking factor, because it is one.
Re-test your PPC strategy against current weights.
Write for AI summary inclusion.
Common mistakes we still see
A few patterns that still show up in accounts optimized for the old playbook:
Title fields stuffed with synonyms and competing brand names, triggering listing quality penalties
Aggressive PPC spend on launches with no plan for organic sustenance, so ranking collapses the moment ad spend pauses
External traffic pointed at raw ASIN URLs without Brand Referral Bonus tagging, leaving the bonus and the attribution data on the table
Inventory planning that optimizes for storage cost over in-stock rate, with predictable ranking damage during demand spikes
A+ Content built as decorative banners with no substantive product information for AI systems to index
Review tactics that violate policy, such as incentivized reviews and review groups, putting the entire account at risk
Conclusion
The “A10” framing is useful shorthand, but it is also a trap. It suggests Amazon’s algorithm is a fixed thing you can decode once and exploit. The reality in 2026 is that Amazon’s ranking system is an AI-mediated layer sitting on top of the same fundamentals that have always mattered: relevance, conversion, customer satisfaction, and account health.
The brands winning on Amazon in 2026 are not doing anything exotic. They are getting the fundamentals right, building listings that AI systems can summarize confidently, and treating Amazon as one channel in a media mix where external traffic, brand strength, and on-platform performance compound on each other. The brands losing are still chasing 2020-era hacks.