Most articles about Amazon ranking factors present a confident numbered list with invented weightings. That confidence is not earned — Amazon has never published a ranking factor list, and the names “A9” and “A10” are community shorthand rather than official products.
What follows separates three things that usually get blended together: what Amazon actually documents, what practitioners have consistently observed, and what is myth. Then it gives you a diagnostic order for a listing that has stopped ranking.
The two layers: relevance, then performance
Amazon search runs as two stages, and almost every seller confusion comes from mixing them up.
| Layer | Question it answers | What influences it |
|---|---|---|
| Relevance / indexation | Should this product appear for this query at all? | Title, bullets, description, backend terms, category and browse node, structured attributes |
| Performance ranking | Of the eligible products, which order? | Conversion rate for the query, sales velocity, click-through, price competitiveness, availability, ratings |
What Amazon actually documents
Amazon's own Seller Central guidance is consistent on a handful of points. These are the safest things to build on:
- Text match matters for discovery. Titles, bullets, and backend search terms are used to match products to customer queries.
- Customer behaviour shapes results. Amazon states that search results reflect what customers click and buy — it is explicitly a behavioural system, not a purely textual one.
- Availability is required. Out-of-stock offers lose placement.
- Complete, accurate attributes help. Structured data (size, colour, material, dimensions) drives filters and refinements, which are a real discovery surface separate from the search box.
- Images and content quality affect conversion, and conversion affects placement.
What practitioners consistently observe
These are inferred from testing at scale rather than documented. They are reliable enough to act on, but hold them more loosely than the list above.
Conversion rate for the query outweighs everything else
Two listings indexed for the same keyword will separate almost entirely on which one converts that traffic. This is why image quality, price, review count, and delivery speed behave like ranking factors even though none of them is a “keyword” thing — they all move conversion.
Title carries the most textual weight
Position within the title appears to matter, with earlier terms weighted more. This is also where keyword stuffing does the most damage: an unreadable title indexes broadly and converts terribly, which is a net loss.
Sales velocity is relative, not absolute
What matters is your velocity compared with competing products for the same query, in the same period. A stable ten units a day is a strong signal in a small niche and irrelevant in a large one.
Stock-outs cost more than they appear to
Going out of stock does not merely pause sales — it surrenders accumulated rank, and recovery usually takes longer than the outage did. This is the single most expensive avoidable mistake in Amazon SEO, and it belongs to operations rather than to copywriting. Our inventory management module exists largely to prevent it.
What is myth
| Claim | Reality |
|---|---|
| Keyword density improves rank | Amazon indexes a term once. Repetition adds nothing and costs readability. |
| Repeating keywords in backend helps | Words already in your title do not need repeating; the space is better spent on synonyms and misspellings. |
| External traffic always boosts rank | It helps only if it converts. Sending cold, untargeted traffic lowers conversion rate and can hurt. |
| Higher price always ranks worse | Price influences conversion, not rank directly. A higher-priced listing that converts better will outrank a cheaper one. |
| There is a fixed factor weighting you can copy | Behaviour differs by category and query. Test on your own catalogue. |
Diagnostic order when a listing stops ranking
Work top to bottom. Do not start rewriting copy until the first four are ruled out — in practice the cause is usually in this half of the list.
- Is it suppressed or inactive? A suppressed listing ranks nowhere. Check this first, always — see fixing suppressed listings.
- Do you still hold the Buy Box? Losing it collapses conversion even though your rank data looks unchanged.
- Did you go out of stock recently? Check the last 30 days, not just today.
- Was the content overwritten? Another contributor to the catalogue can change your title or bullets without warning.
- Are you still indexed for the keyword? Search the exact term plus your ASIN to confirm.
- Did rating or review velocity drop? A slide below 4.0 is usually visible in conversion.
- Did a competitor change price or launch aggressively? Your rank can fall without anything changing on your side.
- Only now, review the copy.
A workflow that compounds
- Find the queries that already convert for you. Your PPC search-term report is the most honest keyword research you own — it is real money against real conversions.
- Move proven converters into the title and bullets, front-loaded and still readable.
- Use backend terms for the leftovers — synonyms, regional variants, common misspellings.
- Fix conversion before chasing more keywords. Images, price, review count, and A+ content move conversion more than any copy edit.
- Protect availability. Everything above is wasted if the SKU goes out of stock.
- Track rank per keyword over time, so you can attribute movement to changes rather than guess.
ListingPilot supports this loop end to end — keyword rank tracking, listing quality scoring, and SEO optimization against live SP-API data. For the copywriting side, see optimising listings with AI; for the ads side that feeds it, see break-even ACoS vs TACoS.