The Amazon ranking algorithm — called A10 by sellers, though Amazon has never blessed the name — ranks products on two things: relevance (does this listing match the query?) and performance (when we show it, do people buy it?). Relevance gets you into the audition. Performance decides whether you keep the part. Everything else you’ve read is a footnote to those two sentences.
That simplicity is also why so much A10 content is nonsense. There’s no leaked factor list, no secret weighting, no ritual involving exactly five bullet points. What there is: a decade of observable behavior, Amazon’s own documentation, and patterns every operator who’s run listings at scale has watched play out. That’s what this post is built from. [Lights the mythology on fire. Warms hands.]
What Is the Amazon Ranking Algorithm?
Amazon’s search ranking system decides which products appear, in what order, for every query typed into the search bar. The original engine was built by A9.com, Amazon’s search subsidiary (whose successor work now lives under Amazon Science) — hence “A9” — and the seller community named its evolved successor A10. Worth saying plainly: A10 is community shorthand, not an Amazon product name. Amazon’s own language, in its Seller Central documentation, is simpler: results rank by how relevant a product is to the query and how well it performs with customers.
The mental model that makes everything else click: Google ranks answers; Amazon ranks predicted purchases. Google can succeed if you read a page and leave. Amazon only makes money when the click converts, so every signal it weighs is a proxy for one question — if we show this product, will this shopper buy it? Optimize for that question and you’re optimizing for the algorithm, whatever letter-number combination anyone calls it this year.
The Relevance Signals: Getting Into the Audition
Relevance signals tell Amazon which queries you belong in. They’re binary-ish — you’re indexed for a term or you aren’t — and they live in the parts of the listing you fully control:
Relevance is table stakes. Finding the terms to target is its own discipline — that’s the keyword research process — but no amount of keyword coverage rescues a listing that fails the next section.
The Performance Signals: Keeping the Part
Once you’re indexed, Amazon starts grading you on behavior, and this is where rank is actually won: click-through rate (of everyone who saw you in results, who clicked?), conversion rate (of those who clicked, who bought?), sales velocity (recent sales weigh more than old ones), price competitiveness (versus the category and versus your own history), in-stock consistency (rank is a habit Amazon breaks quickly when you stock out), and reviews (rating and recency, feeding both CTR and conversion). Notice these compound: better main image lifts CTR, which buys more sessions, which — if the page converts — lifts velocity, which lifts rank, which buys more impressions. That loop is the flywheel everyone names their agency after.
Conversion rate is the kingmaker. Amazon will forgive a mediocre keyword strategy attached to a listing that converts at 25%. It will not forgive a perfectly indexed listing that converts at 6%.
What Changed From A9 to A10?
Directionally — and it’s all directional, since Amazon publishes none of it — the observed shift is from what the listing says toward what shoppers do. Under early A9, keyword coverage could carry a listing surprisingly far. The modern engine leans harder on behavioral signals: organic sales weigh more relative to ad-driven sales, external traffic that converts counts more visibly, and seller-level signals like fulfillment reliability and account health cast a longer shadow. The practical consequence: pure listing-side SEO has diminishing returns, and the conversion side — imagery, price, reviews, A+ Content — is where rank is defended. That’s also why our Amazon SEO service spends as much time on conversion assets as on keywords; the algorithm stopped letting you separate them years ago.
Does External Traffic Help You Rank?
Yes — with a warning label. Orders are orders: a sale that starts on TikTok and ends on your detail page feeds the same velocity signals as one that started in Amazon search, and the branded searches that follow good off-Amazon content (“saw it in a video, typed the brand name”) are among the cleanest ranking fuel there is — that’s the TikTok-to-Amazon flywheel in one sentence. The warning: cold external traffic that clicks and doesn’t buy dilutes your conversion rate, and you already know how the algorithm feels about that. Send audiences that are warm, or send them to a landing experience that filters them first.
Why Did My Rank Drop? (Check These Before Blaming the Algorithm)
When rank falls off a cliff, the algorithm is the last suspect to interview, not the first. The checklist, in order: inventory (low-stock and out-of-stock events kill momentum fastest, and rank rarely comes back for free), Buy Box (lost it and your ads stopped serving properly), price (yours moved, or a competitor’s did), conversion rate (new negative review on top? competitor undercut? main image test gone wrong?), listing health (suppression, category change, stripped content), and only then, genuine algorithm movement. We wrote the full diagnostic sequence in what to do when Amazon sales drop — it exists because “the algorithm changed” is the most expensive wrong diagnosis in ecommerce.
How Long Does Ranking Take? (And the Honeymoon Nobody Tells You About)
New ASINs get a grace period. In the first weeks after launch, Amazon has no behavioral history on your listing, so it extends provisional visibility to find out how shoppers respond — sellers call this the honeymoon phase. It’s the one window where relevance signals carry more weight than performance history, because there is no performance history yet. Launch sloppy — incomplete listing, no reviews plan, no ad support — and you teach the algorithm early that shoppers don’t want you, a lesson that’s expensive to unteach.
For an established ASIN, movement on a given keyword typically shows over weeks, not days, and the speed scales with query volume: you can climb a long-tail term with a handful of conversions, while a head term wants sustained velocity you usually have to buy your way into with ads first. That’s the honest answer behind every “how long does Amazon SEO take” question — it takes as long as it takes you to out-convert whoever currently holds the spot, at a volume Amazon finds convincing.
One more honest note: rank isn’t one number. Amazon personalizes results by shopper history, location, and Prime status, and rank-tracking tools sample one version of a results page that shifts by the hour. Track direction and trend per keyword, not single-day positions — the trend is real; the Tuesday-at-3pm screenshot is weather.
What to Optimize, in Order
If rank is the goal, the order of operations is: fix stock reliability first (nothing survives stockouts), then conversion assets (main image, price position, reviews program), then relevance coverage (title, bullets, backend terms), then paid traffic to prime the velocity loop on terms you can win, then external demand once the page converts well enough to deserve it. Most sellers run this list backwards — keywords first, inventory last — which is how you get a perfectly optimized listing for a product that’s out of stock by Thursday. If you want a second opinion on where your specific listing is leaking rank, that gap analysis is exactly what the first week of a Diagnostic covers.
FAQ
The seller community calls it A10, an evolution of the original A9 engine. Amazon doesn’t use the label publicly — it describes ranking as relevance plus performance: how well a product matches the query, and how well it converts once shown.
Two families: relevance (title, bullets, backend keywords, category) and performance (CTR, conversion rate, sales velocity, price, in-stock rate, reviews). Performance outweighs relevance — a listing that converts poorly slides no matter how well it’s keyword-optimized.
Yes, when it converts. External orders count toward sales velocity, and branded searches that follow off-Amazon content compound the effect. Traffic that clicks without buying drags conversion rate down — send warm audiences, not cold ones.
Check in order: stock position, Buy Box ownership, price vs. competitors, conversion rate, and only then suspect an algorithm update. Most “algorithm” drops are inventory, pricing, or Buy Box events wearing an algorithm costume.