Advertising Amazon Amazon Advertising Amazon Experts Amazon Listing Optimization Amazon Marketplace Amazon News Amazon Prime Amazon Professional Sellers Summit Amazon Seller amazon sellers Amazon Seller Tips Amazon Seller Tools ASIN Brand Management Brands Buy Box Campaign Manager Conference COVID-19 downloadable Dynamic Pricing Ecommerce FBA FBM Holiday Season industry news Multi-Channel Fulfillment Optimize pay-per-click Pricing Algorithm Pricing Software Private Label Profits Repricing Repricing Software Revenue Sales Seller Seller-Fulfilled Prime Seller Performance Metrics SEO SKU Sponsored Products Ads Strategy
Get the latest insights right in your inbox
Published: October 2, 2019
Last updated: March 26, 2026
Marissa Incitti leads research and content at Feedvisor focused on Amazon, Walmart, and the broader e-commerce marketplace ecosystem. Her work covers retail media performance, pricing strategy, and how AI-driven discovery is reshaping how brands compete across marketplaces. Prior to Feedvisor, she worked in content leadership roles at a Fortune Global 500 omnichannel commerce technology company.
Most sellers still optimize their Amazon listings like it’s 2019 - stuffing keywords into every field and crossing their fingers. That approach now actively hurts your ranking. Amazon’s algorithm has shifted so dramatically that the tactics generating visibility three years ago are the same ones suppressing it today.
The core system is still called A9, but what runs under the hood bears little resemblance to the keyword-matching engine it started as. Two AI layers - COSMO for semantic intent and Rufus for conversational shopping - have rewritten the rules. External traffic, which carried zero ranking weight before 2020, now accounts for an estimated 15-20% of ranking contribution. And conversion rate has overtaken raw sales volume as the metric the algorithm cares about most.
Here’s what still drives rankings in 2026 - and the busywork you can stop this week.
Every listing gets scored on two axes. First: does this product match what the buyer typed? The algorithm reads your title, bullets, description, backend search terms, and now - through COSMO - the intent behind the query itself. Second: when buyers land on this listing, do they buy? The algorithm tracks your conversion rate, sales velocity, click-through rate, seller health metrics, and inventory consistency.
You can’t optimize for one and ignore the other. But if you had to pick which to fix first, fix conversion. A 5% conversion rate will bury you regardless of how many keywords you’ve packed in.
The seller community calls the evolved system “A10,” though Amazon has never confirmed that name. What matters isn’t the label - it’s where the weight shifted.
| Factor | A9 (Pre-2020) | Current Algorithm |
|---|---|---|
| Keywords | Exact match dominated | Semantic, intent-based (COSMO) |
| Sales signal | Raw volume | Conversion rate weighted more than volume |
| Traffic sources | Internal only | External traffic rewarded - est. 15-20% of ranking weight |
| Content | Keyword-stuffed listings worked | Natural, benefit-first copy required |
| Reviews | Quantity mattered most | Quality, helpfulness votes, and recency weighted |
| Discovery | Search bar only | Conversational AI (Rufus) creates new channel |
Three shifts deserve special attention.
Average Amazon conversion sits around 9.8%. Top-performing listings hit 15-25%. A product converting at 15% will outrank one with higher total sales but a 5% conversion rate. The algorithm is telling you: it would rather show 10 buyers a listing that 2 of them purchase than show 20 buyers a listing that 1 of them purchases. The math is simple, but most sellers track sales without tracking the rate.
Before 2020, external traffic contributed roughly nothing to ranking. Today, estimate 200-300 monthly external sessions as the minimum for measurable impact; above 1,000 sessions, the boost is equivalent to 10-15 additional daily sales. But here’s the catch: external traffic that bounces tells the algorithm the opposite of what you want. Only converting traffic helps. Sending 500 visitors from a Facebook ad with a 90% bounce rate actively damages your ranking signal.
Seller authority rounds out the picture. Your account health - feedback score, return rate, shipping performance, order defect rate - is now a composite ranking factor. FBA sellers get a built-in advantage because Amazon’s fulfillment guarantees fast shipping and delivery accuracy.
Type “something to keep my coffee hot at the office” and COSMO jumps past keywords to travel mugs - that’s the shift. COSMO is Amazon’s semantic knowledge graph: 6.3 million nodes and 29 million edges connecting products, attributes, and use cases across 18 categories. It doesn’t match your keywords. It infers what the buyer actually wants.
Keyword density now matters less than whether your listing accurately communicates what the product does and who it’s for. Keyword stuffing can harm visibility by signaling poor user experience to the AI. Structure your bullets as feature, benefit, proof, use case - that’s what COSMO reads best.
Rufus is Amazon’s conversational shopping assistant. Over 250 million customers have used it, and shoppers who interact with Rufus are 60% more likely to purchase. Rufus reads your listings, reviews, Q&A sections, and external web content to recommend products. It processes queries like “best gift for a runner under $50” and surfaces products with clear content that answers those questions.
Keywords still matter - augment them with use-case language and conversational phrasing.
Your title is the single strongest relevance signal. Amazon updated title requirements in January 2025:
Front-load your product title with the highest-value keywords. The characters after position 80 are invisible to most buyers - they still get indexed, but they won’t influence click-through rate on mobile.
The backend field holds hidden keywords that are indexed but invisible to buyers. The limit is 250 bytes - exceed it and none of your backend terms get indexed. Not just the excess. All of them.
Reserve this space for synonyms, alternate spellings, Spanish translations (for the US market), and long-tail variations that don’t fit in customer-facing copy. Don’t repeat anything already in your title or bullets - Amazon indexes each unique term only once across all fields, including ASINs and product identifiers (which you should exclude entirely).
Lead with benefits, not features. Address the customer’s problem, then explain how your product solves it. Keep bullets concise - they get truncated on mobile.
For Brand Registry sellers, A+ Content replaces the basic description. Basic A+ lifts sales around 8%; Premium A+ pushes that to 20%. But A+ Content is not directly indexed for keyword search - its value is conversion, not discovery. If your images are weak, A+ Content won’t compensate. Fix images first.
Conversion rate is the metric the algorithm weights most heavily. Average sits at 9.8%. Below 10%, no amount of keyword optimization will save your ranking. Above 15%, you’re in the top tier for most categories.
Sales velocity still matters - estimate a threshold of 25+ sales per month per keyword for meaningful ranking on that term. But sustained daily sales outweigh promotional spikes. A product selling 5 units every day will outrank one that sells 50 in a burst and then goes quiet.
For new products, Amazon PPC campaigns generate the initial velocity needed to enter the ranking conversation. Use PPC search term reports to identify your highest-converting keywords, then work those into your title and backend terms for organic sustainability.
Your price affects conversion rate directly, which affects ranking directly. Static pricing in Amazon’s dynamic marketplace is a slow bleed - competitors adjust in real time, and if you don’t, your conversion erodes. AI-driven price optimization finds the balance between Buy Box eligibility, margin, and sales velocity that manual repricing can’t sustain.
Out-of-stock means zero Buy Box share, zero conversion, and ranking erosion that takes weeks to rebuild. But the less obvious cost is low stock - when you raise prices to conserve remaining inventory, your conversion rate drops, and rankings decay before you actually stock out.
The goal is steady replenishment, especially for your top 20% of selling items. Track inventory against sales velocity weekly.
Reviews are an indirect ranking factor - they drive conversion, which drives ranking. Currently, 94% of US consumers read Amazon reviews before purchasing, and 90% only consider products rated 3 stars or above. The algorithm now weights review quality over quantity: a product with 200 reviews where 80 carry “helpful” votes outperforms one with 500 reviews and few helpful marks.
Bring high-intent traffic and Amazon rewards you; bring bounce traffic and it buries you. Use Amazon Attribution to track which external channels actually convert. Branded search traffic typically converts highest, followed by email, then social - but the spread varies widely by category and audience. Track your own rates in Attribution and cut any channel that consistently runs below half your blended average.
Only send traffic that’s likely to convert. An influencer partnership that drives 300 sessions at a 12% conversion rate is worth more to your ranking than a viral social post that sends 3,000 visitors who bounce.
Your Rankings Are a Pricing Problem Too
Conversion rate drives ranking, and pricing drives conversion. Feedvisor’s AI-powered repricing finds the price point where your sales velocity and margins both win - then adjusts in real time as competition shifts.
See How Feedvisor Optimizes Pricing →A9 is the official name of Amazon’s search algorithm. “A10” is community terminology for the evolved version. In practice, A10 weights conversion rate over raw volume, treats external traffic as a ranking signal (est. 15-20% contribution), factors in seller authority, and uses AI systems (COSMO, Rufus) for semantic intent matching.
Amazon doesn’t provide a native ranking tracker. Third-party tools like Helium 10, Jungle Scout, and SellerSprite track keyword positions over time. Focus on tracking your top 10-15 keywords weekly rather than monitoring hundreds - the long tail moves too much to be actionable.
Not directly. A+ Content is not indexed for keyword search. Its value is conversion - basic A+ lifts sales roughly 8%, Premium up to 20%. Better conversion rates feed back into ranking over time, but if your title and bullets lack the right keywords, A+ Content alone won’t improve discoverability.
25+ per month per keyword gets you into meaningful ranking territory. But conversion rate matters more than absolute volume - the algorithm rewards efficient conversion over brute-force sales numbers.
Rufus recommends products by reading your listings, reviews, and Q&A sections. Products with clear, descriptive content that answers shopping questions get surfaced. Over 250 million customers have used Rufus, and those shoppers are 60% more likely to purchase.
Product Search | Amazon Stores | Seller Performance | Amazon SEO Playbook
Your Rankings Are Slipping - Fix the Algorithm Gaps