Amazon A9 vs A10 Algorithm
What's the difference between Amazon's A9 and A10 ranking algorithms?
A9 (legacy)
Option A- From
- —
- Trial
- —
- For
- —
+ Heavy text-relevance weighting (keywords mattered most)
− Could be gamed by keyword stuffing
A10 (current)
Option B- From
- —
- Trial
- —
- For
- —
+ Behavioral signals dominate (conversion, reviews, returns)
− Harder to optimize without real conversion data
| Dimension | A9 (legacy) | A10 (current) |
|---|---|---|
| Title relevance weight (est.) | ~50% | ~22% (post-May 2026) |
| CVR-14 weight (est.) | Lower priority | ~32% (post-May 2026) |
| Review velocity weight | Lower | ~20% |
| Era | Roughly 2017-2018 | 2018-present |
| Optimization era | Keyword stuffing worked | Behavioral signals dominate |
A9 — the era of keyword density
A9 was Amazon's earlier ranking model. It treated the listing primarily as a text-matching problem: titles, bullets, backend keywords, and the search query were compared, and text-relevance scores drove rank. This era is why so much Amazon SEO advice still focuses on keyword density and 200-character titles — the playbook was built when text features dominated.
A10 — the era of behavioral signals
A10 (or whatever Amazon is internally calling it now) up-weighted behavioral signals: conversion velocity, review velocity, returns rate. Text relevance is still in the model but smaller. The May 2026 rewrite pushed this further: title-keyword relevance lost 6 points; 14-day conversion velocity gained 11. The implication for sellers is that 2018-era advice (stuff every keyword, write 200-character titles) is now actively harmful.