A9, A10, and Why the Name Matters Less Than the Signals
Amazon has never officially confirmed an algorithm called A10. It is community terminology sellers use to describe how the original A9 search engine has evolved, still built on the same keyword-indexing foundation but running far more advanced machine learning on top of it. Arguing about whether A10 is real misses the point.
What matters is that the signals it weighs most heavily today, conversion rate, click-through rate, reviews, return rate, and increasingly seller authority and external traffic, are measurably different from the sales-velocity-first system sellers learned years ago.
The practical takeaway is simple. Optimize for the ranking signals, not for a named algorithm. A listing built around relevance, a strong click-through rate, a high conversion rate, and a genuinely satisfied customer will rank well under whatever Amazon calls its system next.
What Still Moves Rank
Conversion rate is the single most influential ranking factor in Amazon's current system. The percentage of shoppers who click a listing and then actually buy tells Amazon more about product-market fit than any other signal available to it, and it is why two listings with identical traffic can rank completely differently.
- Relevance between keywords and listing content. Title, bullets, backend search terms, and A+ Content still need to genuinely match how real shoppers search, and semantic matching now reads for intent and natural language.
- Click-through rate. The main image, title, and price have to earn the click on the search results page before conversion rate ever comes into play.
- Reviews, rating, and return rate. Review count and star rating remain a strong trust signal, and return rate has become a more direct input, since a high return rate tells Amazon the listing sets the wrong expectations.
- Consistent inventory availability. Going out of stock still costs rank immediately, and that position does not fully return just because stock does.
What Doesn't Move Rank Anymore
- Keyword stuffing. Cramming variations into the title, bullets, or backend terms no longer helps and increasingly hurts, since it produces copy that reads poorly and drags down conversion rate.
- Sales velocity without real conversion quality. Volume generated through discounting, giveaways, or low-intent traffic no longer buys the same rank it used to.
- PPC spend as a standalone ranking lever. Sponsored placements no longer drag organic rank up on their own; PPC works by generating sales history and reviews that feed organic signals.
- Manipulated or incentivized reviews. Detection is sharp enough that review swaps and seeded purchases are more likely to trigger suppression or account risk than to move rank.
- Price as the primary lever. Being cheapest no longer guarantees visibility. A well-converting listing at a fair price consistently outranks a cheap listing with weak content.
What's New in 2026
External traffic from outside Amazon, meaning shoppers arriving through Google, social platforms, or influencer links, now contributes an estimated 15 to 20 percent to ranking outcomes. Amazon rewards products that can prove demand beyond its own search bar, not just within it.
The second development is semantic, AI-driven discovery through systems like COSMO and Rufus, which evaluate listings for natural-language intent and use case rather than exact keyword strings. A listing written in clear, natural, customer-focused language with complete product attributes now has a real advantage over one written purely to satisfy a keyword checklist.
Seller authority, the combined effect of account health, fulfillment reliability, customer service, and consistent sales history, is also treated as a stronger supporting signal than it used to be, even though Amazon has not confirmed it as a formally separate ranking factor.
Ranking Factors at a Glance
| Factor | Status in 2026 | What to Do |
|---|---|---|
| Conversion rate | Highest weight, still rising | Fix listing content, images, and pricing before adding traffic |
| Relevance and natural-language match | Still works, now semantic | Write for real buyer questions, not keyword density |
| Click-through rate | Still works | Invest in main image and title clarity |
| Reviews and return rate | Still works, more weight on returns | Set accurate expectations to reduce returns |
| Inventory availability | Still works, unforgiving | Never let priority SKUs go out of stock |
| External traffic | New, growing weight | Drive qualified traffic from outside Amazon |
| Keyword stuffing | No longer works | Replace with natural, intent-matched copy |
| PPC spend alone | Weaker than assumed | Use PPC to feed conversion data, not as a rank shortcut |
| Sales velocity alone | Weaker than assumed | Pair velocity with genuine conversion quality |
The Outdated Tactics Still Costing Sellers Rank
- Writing titles and bullets as a dense keyword list rather than readable copy, which now works against the listing instead of for it.
- Treating PPC as a rank-buying tool rather than a data and conversion engine, which funds clicks without building the conversion history that protects rank.
- Chasing sales volume through steep discounting without checking whether those buyers convert at a healthy rate or return the product.
- Ignoring content outside Amazon, which leaves the external traffic signal on the table at a moment when it carries real, measurable weight.
