Amazon A9/A10 Algorithm Explained: 2026 Ranking Factors
By Waqar Bajwa, CEO, MBG · Updated October 2026
KEY TAKEAWAYS
- CVR is the highest-weighted A10 signal based on multiple independent correlation studies. Amazon has not published a formal signal-weighting specification, but the pattern holds consistently across thousands of listings tracked by SellerSprite, Signalytics, and Darkroom Agency in 2026.
- A10 weights traffic quality, not traffic volume. A session from branded search earns more ranking credit than an unfiltered social visit to the same page.
- New listings get a temporary ranking window of 30 to 60 days in which velocity signals substitute for missing behavioral history. Sellers who miss this window restart from zero.
- Alexa for Shopping returns one result per query. A listing not in the Featured Offer gets no voice visibility, regardless of its text search position.
- Keyword indexation is the gate in A10: a listing not indexed for a query earns no ranking for that query, regardless of all other signals. Ranking within indexed results is determined by behavior, not keyword density.
Here is what changed. A9 ranked listings that matched queries. A10 ranks listings that convert shoppers. That is not a subtle difference. A listing can match every keyword in a search and still fall below a competitor that converts better, generates more velocity, and pulls qualified external traffic. Most sellers have not updated their optimization approach to reflect this. This guide documents what A10 actually measures, how each signal is weighted based on available industry research, and what that means for how you build and manage listings in 2026.
QUICK ANSWER: WHAT IS THE AMAZON A10 ALGORITHM?
A10 is Amazon’s current ranking system. Where A9 weighted keyword match heavily, A10 weights what shoppers do after they see a listing. Based on SellerSprite, Signalytics, and Darkroom Agency’s 2026 research, the seven primary signals ranked by observed correlation with ranking position are: conversion rate (highest), sales velocity, click-through rate, seller authority, external traffic quality, review velocity, and keyword relevance (lowest). Amazon has not published a formal ranking specification; these weights reflect observed correlation, not a confirmed internal model. In practice, a listing converting 15% with average keyword coverage typically outranks a listing with perfect keyword coverage converting 9%. Keyword optimization is still required for indexation; it no longer determines rank.
A9 vs. A10: What Actually Changed
The practical difference between A9 and A10 is the shift from input signals to output signals. A9 rewarded what you put into a listing (keywords, categories, backend terms). A10 rewards what shoppers do when they find it. The same listing, optimized exactly the same way, ranks differently under A10 because the behavioral feedback loop now drives the outcome.
| Ranking Signal | A9 Behavior | A10 Behavior | Practical Impact |
|---|---|---|---|
| Keyword match | Primary ranking signal | Required for indexation only | Keywords open the door; behavior determines rank |
| Conversion rate (CVR) | Secondary signal | Highest-weighted factor (by observed correlation) | CVR gap between competitors now determines position |
| Sales velocity | Important | High weight, direction-sensitive | Rising velocity earns a temporary rank boost |
| External traffic | Not differentiated | Quality-weighted by behavior | High-intent referral traffic outperforms volume traffic |
| Seller authority | Minimal influence | Baseline multiplier, medium weight | Same listing ranks higher on a high-authority account |
| Review signals | Volume-weighted | Velocity and recency weighted | Fresh reviews outrank larger but older review sets |
| Click-through rate | Secondary | Tracked separately from CVR | High CTR + low CVR can suppress ranking on buy queries |
| Semantic coverage | Keyword-based | Entity-based via COSMO model | Use-case entities in copy and reviews improve indexation |
Sources: SellerSprite Algorithm Analysis 2026; Signalytics Ranking Studies 2026; Darkroom Agency A10 Research 2026
The 7 A10 Ranking Factors, Ranked by Weight
Amazon has not published a formal ranking specification. The weight order below is based on SellerSprite’s 2026 algorithm analysis, Signalytics ranking correlation studies across thousands of listings, and Darkroom Agency’s longitudinal testing. No external source has direct access to Amazon’s model; the estimates below reflect observed correlation between each signal and ranking movement, not confirmed internal weights.
| Ranking Factor | Estimated Weight | Primary Control Point |
|---|---|---|
| 1. Conversion rate (CVR) | Highest | Main image, price, reviews, A+ Content |
| 2. Sales velocity | High | PPC spend, promotions, inventory |
| 3. Click-through rate (CTR) | High | Main image, title, star rating, price |
| 4. Seller authority | Medium | FBA enrollment, account health, in-stock rate |
| 5. External traffic quality | Medium | Branded channels, Attribution tag tracking |
| 6. Review quality and velocity | Medium | Vine, post-purchase follow-up, product quality |
| 7. Keyword relevance | Lower (not zero) | Title placement, backend search terms (249 bytes) |
Factor 1: Conversion Rate (CVR): Highest Weight
What A10 measures
The percentage of product page sessions that result in a purchase, compared against the median CVR for the listing’s subcategory and keyword cluster. A10 does not apply a single global CVR benchmark. A listing in the Supplements subcategory is measured against Supplements norms; a Kitchenware listing is measured against Kitchenware norms. This matters because a 12% CVR is above median in some categories and well below in others.
The benchmark
SellerSprite’s 2026 correlation study across competitive categories found that listings in the top 3 positions averaged approximately 14% CVR, against under 7% for listings outside the top 10. The CVR gap was more strongly correlated with position than keyword density or review count. The implication is that closing a CVR gap of 3 to 5 percentage points typically has more ranking impact than adding 50 new keywords to backend search terms.
What sellers can control
Main image quality is the single largest CVR lever in most categories because it drives both the click decision and the purchase decision. Price competitiveness relative to visible competitors on the same search page is the second lever. Review count and recency are third. A+ Content layout and clarity improve CVR for mid-consideration shoppers who are close to purchasing but not yet convinced. Title specificity that sets accurate expectations reduces returns and improves post-purchase review sentiment, which feeds back into future CVR.
→ Listing Optimization Service
Factor 2: Sales Velocity: High Weight
What A10 measures
Daily and weekly unit volume relative to category benchmarks, plus the direction of that velocity over time. A10 distinguishes between a listing generating 10 units per day with rising trajectory and the same listing generating 10 units per day with a falling trajectory. The two rank differently. A rising velocity listing receives what the industry informally calls the new entrant advantage: a temporary ranking boost that sustains as long as velocity continues to rise.
The benchmark
Signalytics’ 2026 data found that new listings reaching 3 to 5 daily organic units within the first 30 days retained ranking momentum significantly longer than listings remaining at 0 to 1 daily organic units. The threshold is not a universal rule: it varies by category and subcategory. The principle holds broadly: advertising investment during launch is not optional if sustained organic ranking is the goal.
What sellers can control
PPC spend at launch is the fastest direct velocity input because it generates sessions and purchases immediately. Coupons and promotional pricing contribute to velocity but at a lower quality signal than organic sales. Inventory availability is critical: a stockout resets velocity signals and is difficult to recover from. Based on MBG’s experience managing Amazon brands through Q4 cycles, sellers who go out of stock during a rising velocity window typically need weeks to months to rebuild ranking gains, depending on category competition and how long the stockout lasted.
Factor 3: Click-Through Rate (CTR): High Weight
What A10 measures
The ratio of clicks to impressions in search results, tracked as a distinct signal from CVR. A listing with 8% CTR and 15% CVR tells the algorithm that a specific audience finds and buys this product. A listing with 15% CTR and 6% CVR tells it that shoppers are drawn in but not convinced. These two signals are processed separately, and the combination determines whether the algorithm surfaces a listing for high-purchase-intent queries or broad informational queries.
The benchmark
Main image testing consistently produces the largest CTR improvements. Based on MBG’s main image testing across client accounts in 2025 and 2026 in categories including Home, Health, and Sports, image swaps produced CTR improvements of 20 to 40% within 30 days. Title changes that reduce ambiguity or better match the dominant query intent add smaller but additive CTR effects. Star rating display affects CTR more than most sellers account for: listings below 3.8 stars show substantially lower CTR regardless of image quality or price.
What sellers can control
Main image is the highest-leverage CTR variable. Title clarity is second: a title that tells the right shopper exactly what the product does generates a more qualified click than a title padded with keywords. Badge eligibility (Best Seller, Amazon’s Choice, Climate Pledge Friendly) improves CTR visibility. Price display against visible competitors on the same results page affects CTR directly. These inputs compound: improving three of them simultaneously typically produces a larger CTR lift than doubling down on one.
Factor 4: Seller Authority: Medium Weight
What A10 measures
A composite account-level score derived from fulfillment rate, account health dashboard metrics, FBA utilization, return rate, and customer service response time. Seller authority acts as a baseline multiplier applied to ranking signals across all of a seller’s listings. An identical listing on a high-authority account ranks higher than the same listing on a lower-authority account because the algorithm weights its behavioral signals more heavily.
What sellers can control
FBA enrollment is the single most impactful seller authority decision because it improves the score directly and simultaneously improves Buy Box eligibility, which affects Alexa for Shopping visibility. Maintaining an Order Defect Rate below 1%, responding to customer messages within 24 hours, and keeping in-stock rates above 95% protect the baseline score. Seller authority cannot be bought or accelerated. It accumulates from consistent performance over months. The only way to damage it quickly is through fulfillment failures, account health warnings, or a sustained period of out-of-stock.
Factor 5: External Traffic Quality: Medium Weight
What A10 measures
Not all external traffic earns the same ranking credit. A10 weights external sessions by their behavioral outcome on Amazon. A visitor arriving from a branded Google search who views the product for 90 seconds and purchases earns substantially more ranking credit than a visitor from a broad TikTok ad who bounces in 12 seconds. Amazon’s Brand Referral Bonus (averaging approximately 10% credit on qualifying sales, per Amazon’s official program documentation) is a direct signal that Amazon actively rewards sellers who drive high-quality external traffic.
What sellers can control
The channel determines quality. Branded search campaigns (Google, Bing), direct email to past customers, editorial press coverage, and organic social from engaged audiences produce higher-quality behavioral sessions than undifferentiated paid social. Amazon Attribution tags let sellers measure which external channels produce sessions that actually convert on Amazon. Sellers using Attribution can identify channels generating below-baseline CVR and reallocate budget toward those generating above-baseline CVR. Low-quality traffic that fails to convert actively suppresses CVR signal and can hurt ranking net of the traffic volume benefit.
Factor 6: Review Quality and Velocity: Medium Weight
What A10 measures
The rate at which new verified reviews arrive (velocity) and the informational content of review text (quality). A10 does not weight a 4-year-old 5-star review the same as a 4-week-old 5-star review. Recency matters. Amazon’s COSMO semantic model also reads review text for entity signals: a review mentioning “waterproof for hiking” helps COSMO associate a listing with outdoor use-case queries that the listing’s own copy may not explicitly target. This means review vocabulary influences indexation, not just star rating.
What sellers can control
Amazon Vine seeds high-quality reviews from verified reviewers and typically produces more detailed text with images, which strengthens COSMO entity signals. Post-purchase email follow-up sequences (compliant with Amazon’s messaging policies) accelerate velocity. Product quality at scale drives organic review volume over time. Targeting specific use cases in packaging and inserts can influence the vocabulary of incoming reviews, which in turn improves COSMO entity matching for those use-case queries.
→ Listing Optimization Service
Factor 7: Keyword Relevance: Lower (Not Zero)
What A10 measures
Whether a listing is indexed for a given query at all. Keyword relevance in A10 is a gate, not a ranking driver. Once a listing is indexed for a query, keyword density stops influencing where it ranks. If a listing is not indexed for a query, it cannot appear in results for that query regardless of how well it would convert.
What sellers can control
Title placement carries the most indexation weight. Backend search terms (249 bytes, use them fully) index terms that do not fit naturally in visible copy without making copy awkward. Category and browse node selection affects indexation for category-based queries. Keyword cannibalization between listings on the same account (two listings competing for the same query) can reduce indexation depth for both. Running an indexation audit before starting any other optimization identifies which ranking deficits are structurally blocked versus underperforming. A listing not indexed for its primary query cannot improve its CVR signal for that query regardless of how much the listing itself improves.
QUICK ANSWER: HOW DOES ALEXA FOR SHOPPING AFFECT A10 RANKINGS?
Alexa for Shopping returns exactly one product per query: the current Featured Offer. A listing not in the Featured Offer position receives zero voice search visibility. The Featured Offer is determined by Buy Box eligibility first (FBA sellers with healthy accounts win the Buy Box) and then by A10’s standard ranking signals in the same order they apply to text search. Optimizing for A10 text search ranking and optimizing for Alexa for Shopping are the same task. There is no separate voice search optimization strategy.
New Listing vs. Mature Listing: Different Priorities
The same seven A10 factors apply across all listings, but their relative priority shifts depending on how much behavioral history a listing has accumulated. A new listing has no CVR data for the algorithm to weigh, so it defaults to velocity signals to estimate ranking position. A mature listing with 6+ months of history is ranked almost entirely on behavioral outputs.
| Lifecycle Stage | Primary Ranking Lever | Secondary Lever | The Risk of Getting It Wrong |
|---|---|---|---|
| New listing (0 to 30 days) | Sales velocity (advertising-driven) | Review acquisition | Missing the launch window forces a full restart; velocity signals reset on stockout |
| Growth phase (30 to 180 days) | CVR improvement | CTR optimization | Continuing to spend on velocity without fixing CVR compounds the loss; every new session converts poorly |
| Mature listing (180+ days) | CVR sustained above category median | External traffic quality | Treating a mature listing like a launch wastes spend on velocity that the algorithm no longer weights as heavily |
“A brand-new listing has no conversion history. The algorithm has nothing to rank it on. Sales velocity is the only signal you can manufacture at launch. Once you have 30 to 50 reviews and a conversion rate above 10%, the priority shifts entirely to CVR and external traffic. Treating a mature listing like a launch listing is one of the most common ranking mistakes we see.”
Waqar Bajwa, CEO, MBG
Alexa for Shopping: The Third Ranking Surface
Amazon’s search ecosystem operates across three surfaces: desktop and mobile text search, mobile app search, and Alexa for Shopping. Alexa for Shopping mediates an estimated 15 to 20% of Amazon mobile queries (Signalytics, 2026). It returns one result per query. Any listing not in the Featured Offer position receives zero voice search visibility.
Three things determine whether a listing can appear in Alexa results:
- Buy Box eligibility (required; FBA with healthy account health is the most reliable path)
- A star rating sufficient to clear Amazon’s voice search filter (Signalytics’ testing indicates listings below 3.8 stars are frequently excluded from voice results, though Amazon has not confirmed a specific threshold)
- A10 ranking position (Alexa returns the listing A10 already ranks first; there is no separate voice ranking layer)
The COSMO semantic model connects voice queries to listings through entity matching rather than keyword matching. A shopper asking Alexa for “a protein powder that doesn’t clump” is not triggering a keyword search for “no-clump protein powder.” COSMO reads use-case entities in listing copy and reviews and maps them to natural-language queries. Listings with rich entity coverage (use cases, materials, compatible items, outcomes) index better across voice queries than listings optimized only for exact-match keyword coverage.
Practical optimization steps for Alexa visibility:
- Maintain FBA enrollment and Buy Box eligibility as the non-negotiable prerequisite
- Keep star rating above 3.8 through active review management and product quality
- Use natural-language entity coverage in bullets and A+ Content to improve COSMO matching
- Target conversational use-case queries in backend search terms (these tend to match Alexa query patterns)
Frequently Asked Questions
Amazon has not confirmed A10 exists. Is this framework actually reliable?
Amazon has never publicly labeled its algorithm “A10.” The term was coined by industry analysts who observed a consistent behavioral shift beginning around 2021 and accelerating through 2023 and 2024. Whether the internal label is accurate is beside the point: the behavioral signals the framework describes are confirmed by correlation studies across thousands of listings from multiple independent sources. The optimization implications hold regardless of what Amazon calls the system internally.
If CVR is the most important signal, why does keyword research still matter?
Keywords determine whether a listing indexes for a query at all. A listing converting 20% that is not indexed for “wireless earbuds under $50” earns zero ranking for that query. Keyword research identifies which queries to index for; CVR determines where the listing ranks once indexed. Both are required; neither substitutes for the other. The common mistake is spending optimization time on keyword density after indexation is already adequate, rather than addressing the CVR gap that is actually limiting rank.
How do I know whether my external traffic is helping or hurting my A10 ranking?
Use Amazon Attribution tags on all external traffic links. The Attribution dashboard shows which campaigns produce sessions that convert on Amazon and which produce sessions that bounce. If an external channel is generating a conversion rate below your listing’s organic baseline, it is suppressing your CVR signal. If it converts above baseline, it is contributing a positive signal. Cut channels showing consistently below-baseline CVR; increase spend on channels showing above-baseline CVR. Volume alone is not the objective.
My listing has strong keywords and reviews but rankings keep dropping. What is A10 checking?
Dropping rankings despite adequate keyword coverage and reviews typically point to a CVR or velocity problem relative to competitors, not a problem with the listing itself. A10 recalibrates continuously. If a competitor improved its main image and raised CVR from 9% to 14%, your position falls without any change on your listing. Run a CVR comparison against current page-1 competitors, check your Units Ordered trend in Seller Central for a velocity decline, and compare your main image against what is currently visible on the search results page for your primary keyword. In our experience, those three checks resolve the large majority of cases where rankings drop without any change to the listing itself.
Does COSMO affect all product categories the same way?
No. COSMO’s influence is strongest in categories where queries use natural language and use-case phrasing rather than exact product specifications: Health, Pet, Sports, Outdoor, Beauty. In categories dominated by model numbers and technical specifications (Electronics, Automotive Parts, Industrial), keyword match remains the primary indexation mechanism because shoppers search with precision. COSMO becomes more relevant as query language becomes more conversational and intent-based, which is also where Alexa for Shopping query patterns are most concentrated.
Your A10 Priority Order: What to Do First
Understanding the seven signals is useful. Knowing the order to address them is what actually moves rankings. The steps below apply to any listing; the lifecycle stage determines where you start.
- Step 1: Run an indexation audit. Confirm your listing indexes for its primary and secondary keywords. A listing not indexed for a query earns zero ranking for that query regardless of CVR. Fix indexation failures first. Everything else builds on this.
- Step 2: Compare your CVR against visible page-1 competitors. Pull your CVR from Seller Central Business Reports and compare it to the category median. If your CVR is below the median for your subcategory, this is the ranking bottleneck, not keywords. Fix main image, price, and review presentation before touching backend search terms.
- Step 3: Identify your lifecycle stage and match your velocity strategy. New listing (0 to 30 days): invest in PPC to build velocity before the launch window closes. Growth phase (30 to 180 days): shift spend toward CVR and CTR improvement. Mature listing (180+ days): protect CVR above the category median and invest in branded external traffic channels.
- Step 4: Address CTR if CVR is already above median but ranking lags. High CTR with low CVR suppresses ranking on purchase-intent queries. Low CTR with high CVR limits the volume of sessions the algorithm can evaluate. If CVR is strong but position is stuck, compare your main image and title against current page-1 results for your primary keyword. Test a new main image before changing any other variable.
- Step 5: Protect seller authority as the baseline. Enroll in FBA if not already. Keep in-stock rates above 95%. Respond to customer messages within 24 hours. Monitor your Account Health Dashboard weekly. Seller authority cannot be improved quickly, but it can be damaged quickly, and a damaged baseline suppresses all ranking signals simultaneously.
Common mistake: Sellers jump to keyword optimization (Step 1) and stop there. If the listing is already indexed, adding more keywords to backend search terms produces no ranking improvement. The issue is always CVR, velocity, or CTR. Keyword work is only the first diagnostic, not the fix.
The most common mistake sellers make in 2026 is treating A10 as a more complex version of A9. It is not. A9 was an optimization problem: put the right terms in the right places. A10 is a performance problem: get the right shoppers to convert at a rate above your category median. Keyword optimization still matters, but only as the prerequisite that gets a listing into the competition. What happens after the click determines the ranking outcome. Sellers who understand that distinction stop optimizing for visibility and start optimizing for conversion. The ranking follows.
GET A FREE AMAZON LISTING AND ALGORITHM AUDIT
MBG identifies your A10 ranking gaps across CVR, velocity, indexation, and seller authority.
Related: TACoS vs ACoS: The Metric Amazon Sellers Get Wrong →