The A9 algorithm is Amazon's search-ranking system, distinct from Cosmo (Amazon's Rufus-Powering Algorithm) which powers Rufus. A9's stated aim is to maximize Amazon's revenue by surfacing products shoppers are most likely to buy, weighing keyword relevancy and sales performance (conversion rate, sales velocity).
Relevancy is the lever sellers fully control, versus sales performance which depends on market demand and price — so keyword placement work under Amazon Listing Optimization: SEO vs. Conversion Optimization targets A9 directly: get the keyword into the listing so A9 can match it, then win the click-to-sale conversion so A9 rewards the listing with rank.
The 3-Step SEO Framework (Keyword Research → Build → Validate) (keyword research → listing build → indexing/rank validation) exists because A9 only ranks on keywords it can see and that are converting.
A9's ranking of search results is driven by two main inputs: keyword relevancy (is the listing indexed for, and does its content match, the searched term) and sales performance (does the listing actually convert and sell once it appears). The algorithm's single overriding goal is commercial: "Amazon's A9 algorithm, which is the brain behind the Amazon search, has one main goal, and that's to make Amazon money" — it surfaces listings likely to complete a transaction, not listings that are merely topically relevant. This is the underlying reason PPC and launch pricing exist as separate levers: a brand-new listing has no sales performance yet to rank on, so it needs paid visibility and price-driven conversion to accumulate the sales signal A9 rewards.
The top organic search results for a given term function, in effect, as a sales leaderboard: the #1 organic listing for a search is typically the best seller for that search, not simply the listing A9 judges most "relevant." Ranking is a trailing indicator of sales performance for that keyword, which is why converting well on a keyword compounds into better organic placement over time.
A9 computes ranking, per keyword, as performance × relevancy.
Because Amazon can't wait for a distinct click or purchase against every individual keyword a listing could match, every click or purchase gives partial "broad" performance credit across many related keywords at once. That credit is only "unlocked" for a given keyword in proportion to that keyword's relevancy score — so a listing with high performance but low relevancy on a term still won't rank well for it, and vice versa.
Rank can be modeled as a function of performance multiplied by relevancy, calculated per keyword — not performance or relevancy alone. A keyword where a listing is highly relevant but rarely converts, or converts well but is barely relevant, underperforms one where both factors are strong.
Apply: to move rank on a target keyword, raise both sides together — relevancy (see A9 Title Keyword Weighting (Position & Phrase Match) for exact-match placement) and performance (CTR, conversion rate, revenue on that keyword) — since optimizing only one half of the multiplication leaves rank gains on the table. This is also why the practical launch order runs relevancy first (write the listing with exact-match keywords), then performance (validate conversion), then turn on PPC with the right keywords already in place.
Из тем: Unsorted, SEO & Keyword Strategy: Winning A9, Cosmo & Rufus