What is organic rank on Amazon?

Organic rank is a product's position in Amazon's unpaid search results for one specific query. It is distinct from Best Sellers Rank, which measures category sales volume with no query attached; from the Featured Offer, which decides which offer wins the add-to-cart button; and from what Amazon's AI shopping assistant recommends. A product can rank well in one of those systems and be invisible in another.

How do I improve my Amazon ranking?

Diagnose before you act. Amazon discovery has four jobs — eligibility, retrieval, selection and conversion — and each fails differently. Eligibility failures produce no impressions at all, so no ranking report will show them. Retrieval failures mean the product is not indexed for the query, so no bid or copy change helps. Only once the failing layer is named does the choice of tactic matter.

Does Amazon publish its ranking factors?

No. Amazon's official SEO guidance makes one causal statement — that listing quality and account health can contribute to search rankings — and publishes no factor list, no weights and no thresholds. The best-documented description of the system remains a 2016 SIGIR paper by the A9 team, which establishes that behavioural signals such as clicks and purchases drive Amazon rankings far more than they drive web search.

The Amazon discovery framework · v3.2 · 22 August 2026

Amazon ranking, without the folklore.

The complete operating model for Amazon organic discovery: the layer stack from lexical matching to agentic commerce, the measurement system, and the diagnostics that name which layer is actually failing. Every claim labelled by where it came from.

How to rank on Amazon Score my listings

Independent publication. Not affiliated with, endorsed by or sponsored by Amazon.com, Inc. Trademark notice & disclaimers.

If you are working from anything published before August 2026

Titles were cut to 75 characters on 27 July. Item Highlights is locked until they are. Review sharing across variations ended. Featured Offer eligibility requirements were removed. Rufus became Alexa for Shopping. Every one of those invalidates advice that was correct last year. The dated changelog →

Most optimisation fails at the diagnosis, not the execution

Bidding on a query the ASIN is not indexed for. Rewriting copy when the offer is the constraint. Reworking images when an attribute silently reverted three weeks ago. The tactic is usually fine. The layer was never named.

Amazon discovery has four jobs, and every optimisation decision belongs to exactly one of them. Naming the job before acting is the whole discipline.

1 · Eligibility

Can the ASIN enter the candidate set at all? Attributes, product type, browse classification, identifier mapping, variation validity. A failure here is invisible in every ranking report you own — the ASIN is simply not considered.

2 · Retrieval

Can Amazon find it for the shopper’s language? Indexation across name, highlights, bullets, description and backend — lexical and semantic.

3 · Selection

Is it a strong fit for this shopper in this context? Evidence quality, behavioural association, personalisation, the assistant layer.

4 · Conversion

Does the tile, the page and the offer win? Price, delivery promise, reviews, images, stock.

The diagnostic discipline that follows

If you are not indexed, PPC cannot rescue you. If you are indexed but deep, copy may not be the bottleneck. If you are on page one but never surfaced by an assistant, evidence is the missing layer. Name the layer before acting.

How this site handles evidence

Amazon publishes very little about how ranking works. That absence is usually filled with confident invention: weighted factor lists, percentages, version numbers for algorithms that do not exist. We fill it with labels instead.

Amazon confirmed Amazon’s own documentation

Amazon research Published Amazon research

Observed Reproducible practitioner testing

Model This framework’s inference

The rule is simple and it is enforced on every page. Never present an observed pattern as confirmed architecture. Never use an unsourced precise number as a planning input. There is no A10, A11 or A12 — there has only ever been A9, evolving in layers, and even that name is now an anachronism.

Start here

The framework

The guide

How to rank on Amazon

The complete method. The four jobs, the ranking equation, the diagnostic order of operations, and a sequenced plan for week one, month one and quarter one.

Architecture

The six-layer discovery stack

Lexical, semantic, behavioural, the knowledge graph, the assistant, agentic commerce. No layer replaced the one before it, and each inherits the failures below.

Eligibility

The catalog layer nobody audits

Product type, contribution authority, the three-state attribute census, variation architecture. The layer where you can be wrong for months while every dashboard agrees with you.

Construction

The 75/125 title system

The full specification for the July 2026 title change, the Highlights lock, the placement hierarchy, and a migration sequence that does not lose queries.

Relevance

Being quotable: the assistant layer

Retrieval is the eligibility gate for the entire AI layer. Then extractability decides whether anything you wrote can actually be used.

Measurement

The four-index diagnosis

Search Query Performance read correctly: the mandatory splits, the four indices from raw counts, the significance floors, and why the percentage columns invert diagnoses.

Strategy

Keyword strategy

The query universe from first-party data, six classification dimensions, the promotion test, the launch set and the V×R×W push ladder.

Lifecycle

Launch, day 0 to day 90

Eligibility before relevance, relevance before conversion, conversion before spend. Plus the failure signatures, and how to diagnose decline honestly.

Spend

PPC as a ranking instrument

Spend does not buy rank. It buys the behaviour that produces rank, and only above the 0.9 purchase-index gate. Placement economics and TACOS as a budget.

Incidents

Scenario runbooks

Rank dropped. Buy Box lost. Listing suppressed. Ads spending, nothing ranking. Incident inside a deal window. Check sequences and decision points.

“Amazon ranking” is four different systems

The second most common error, after skipping the diagnosis, is treating Amazon visibility as one ranking with one set of rules. It is at least four, with overlapping but genuinely different inputs.

System	What it ranks	What actually drives it
Organic keyword rank	Your position for one specific query	Retrieval eligibility, then behavioural performance on that query
Featured Offer	Which offer wins the add-to-cart button	Competitive pricing, delivery speed, performance metrics Amazon confirmed
Best Sellers Rank	Sales volume relative to a category	Recent and all-time sales; explicitly not reviews or page views Amazon confirmed
Assistant recommendation	What Alexa for Shopping surfaces conversationally	Extractable evidence, and being in the candidate pool at all

Observed A study of more than 1,300 AI-assistant recommendations found a median organic search rank of 41 among recommended products, with the 75th percentile not ranking in organic search at all. Products were winning one visibility system while invisible in another. Treating them as one thing is the root error in most competing content. The full stack →

Diagnose something specific

Most common

My organic rank dropped

Nine causes, ordered by how often they turn out to be the real one, with the check that distinguishes each.

Retrieval

My product is not indexed

How to check indexation properly, the five reasons it fails, and why indexed is not the same as ranking.

Tool

Listing readiness scorecard

Twenty checks across the five layers, scored 0–100, returning a ranked issue list by severity. No signup to see your result.

Tool

Title & backend byte checker

Counts the 75/125 fields in code and the backend field in bytes. Flags the Highlights lock. Runs in your browser.

Tool

SQP four-index calculator

Raw counts in, the failing stage named, the significance floors applied, and the spend gate checked.

Correction

A9, A10, COSMO, Alexa

One of those four names refers to something that never existed, and it is the one you have read the most about.

Reference

Glossary

Organic rank, indexing, BSR, rank juice, sales velocity, cold start, COSMO, SQP. Sourced, and honest about what nobody knows.

The framework is free. Running it across a catalogue is the job.

Twelve years of seller and agency operations went into this. If you would rather have it applied to your account than explained to you, the teardown is free, takes 72 hours, and the findings are yours to keep either way.

Get a free account teardown →