Amazon SEO: How the A9/A10 Algorithm Ranks Products

Amazon's search engine is not Google. It has one objective, which is to show the product most likely to result in a purchase, and it has direct data on what shoppers buy after searching. That makes Amazon SEO simpler in principle and harder in practice: the words on your listing get you into the results, and the shopping behavior on your listing determines where you sit in them.

A9 was the name of the Amazon subsidiary that built product search, and A10 is an industry label for the later evolution rather than an official Amazon term. The labels matter less than the mechanics, which have been stable for years: a relevance filter, then a performance ranking. This guide explains both, how to research and verify keywords, and which shortcuts do more harm than good.

Relevance signals: getting into the results

The first thing the algorithm does with a search is find every product whose indexed text matches the query. Indexed text comes from the title, bullet points, description, backend search terms, and structured attributes such as brand, color, material, and product type. If a term does not appear in any of those fields, your product does not appear for that search, regardless of how well it sells.

Field weight is not equal. The title carries the most, bullets and structured attributes next, then the description and backend terms. Word order and proximity matter less than they once did; Amazon matches individual terms and combinations across fields, so "stainless steel water bottle" can match a listing with "water bottle" in the title and "stainless steel" in a bullet, though a listing with the full phrase in the title will typically be preferred for that exact query.

Category placement is also a relevance signal. A product listed in the wrong browse node can be indexed for a term and still fail to appear because shoppers filtering by category never see it. Choose the most specific node that fits, and verify the item type attribute matches.

Performance signals: where you rank

Among relevant products, ranking is driven by performance: the probability that showing your product results in a sale. Amazon estimates that from your history on each query and overall.

The signals are the ones you would expect. Click-through rate from search results, driven by the main image, title, price, review count and rating, and Prime eligibility. Conversion rate once the shopper is on the page, driven by everything on the listing plus price and availability. Sales velocity, meaning recent unit sales and their trend, weighted toward the specific search terms that produced them. Returns and negative feedback pull in the other direction, since they indicate a sale that did not stick.

There are also gating factors. A listing that is out of stock drops from ranking quickly and does not immediately recover when restocked. A listing without the featured offer (the Buy Box) converts poorly and ranks accordingly. A listing suppressed for a policy issue, such as a title over 200 characters or a main image with text on it, disappears entirely.

Keyword research methods that work on Amazon

Amazon keyword research starts with Amazon's own data. Sellers enrolled in Brand Registry have Brand Analytics, which includes Search Query Performance (the actual queries that led to impressions, clicks, and purchases for your brand) and the Search Terms report (top queries across Amazon with the three most-clicked ASINs for each). Those two reports alone will tell you which terms drive your category.

Without Brand Registry, the search bar's autocomplete suggestions are a reliable window into real query patterns. Type your product type and note the suggestions, then repeat with each modifier. Third-party keyword tools estimate volume and show which terms competitors are indexed for; treat their volumes as relative, not exact.

Sponsored Products campaigns are the most direct research method available. Run an automatic campaign for two to three weeks, then pull the Search Term report to see which queries actually converted. Those terms belong in your title and bullets.

• Brand Analytics: Search Query Performance and Search Terms reports, if registered.
• Autocomplete: product type plus each modifier (material, size, use case, audience).
• Competitor listings: titles and bullets of the top three ranked products for your main term.
• PPC Search Term report: the queries that converted, not just the ones that got clicks.
• Customer reviews and Q&A: the words shoppers use to describe problems and uses.

How to check whether you are indexed

Being indexed for a term means Amazon will consider your product for that search. The simplest check is to search for your ASIN followed by the term, for example "B0XXXXXXXX ceramic pour over dripper." If your product appears, you are indexed for that phrase. If nothing appears, the term is missing from your indexed fields or the listing is suppressed.

Do this for your top 20 to 30 terms after every listing change, and pay attention to individual words within phrases. A product can be indexed for "pour over dripper" but not "ceramic" if that word appears only in an image. Some keyword tools automate the check and report indexed status per term.

Also verify that your structured attributes are populated. Amazon increasingly relies on attributes for filters and matching, and a listing with material, color, and size left blank loses indexing that no amount of keyword placement recovers.

Sales velocity and conversion: the loop

Ranking on Amazon is a feedback loop. Better rank produces more sessions, more sessions with a strong conversion rate produce more sales, and more sales on a term improve rank for that term. The loop runs in reverse too, which is why a stock-out or a price increase can cost a position that took months to earn.

Two practical consequences follow. First, conversion rate is the lever most in your control: every listing improvement that raises conversion improves ranking without spending on ads. Second, launches need an initial push, since a new listing has no sales history for the algorithm to evaluate. That push usually comes from Sponsored Products targeting the exact terms you want to rank for, priced to win impressions while the listing gathers its first sales and reviews.

Track sessions, unit session percentage (Amazon's name for conversion rate), and ordered units per ASIN in Business Reports. When a change to the listing improves unit session percentage, you should see organic rank follow within a few weeks.

External traffic and what it does

Traffic from outside Amazon, such as your own site, email list, social channels, or influencers, counts toward sales velocity like any other sale, and Amazon actively encourages it. Brand Registered sellers can use Amazon Attribution to tag external links and see which sources convert, and the Brand Referral Bonus program credits back a portion of the referral fee on qualifying sales that originated off Amazon.

The caveat is conversion. External traffic that does not convert lowers your unit session percentage, which can hurt more than the extra sales help. Send warm traffic, such as an email to existing customers or a targeted post, rather than cold traffic, and land it on a listing that is ready to convert. Some sellers land external traffic on a Brand Store page rather than the product page for exactly that reason.

What not to do

Amazon's rules are enforced by both algorithm and policy team, and the penalties range from lost indexing to listing suppression to account suspension. The tactics below are common, tempting, and counterproductive.

• Keyword stuffing in the title: the 2025 title rules cap repeated words at two occurrences, and a cluttered title lowers click-through, which hurts ranking.
• Competitor brand names in backend search terms: a policy violation that can suppress the listing.
• Incentivized reviews, review swaps, or asking customers to leave a positive review: prohibited since 2016, with enforcement that includes review removal and account action.
• Manipulating sales velocity through rebate schemes or search-find-buy campaigns: Amazon's Seller Code of Conduct treats these as ranking manipulation.
• Merging unrelated products into one variation family to share reviews.
• Frequent listing edits during a launch: each edit can trigger re-indexing, so batch changes and give them time to take effect.
• Ignoring stock: running out of inventory resets ranking momentum.