Amazon A9 Algorithm Factors: The Complete Ranking Framework

Amazon's A9 algorithm determines which products appear at the top of search results — and which disappear into page 10 obscurity. Most articles on A9 ranking factors recycle the same generic advice or repeat outdated assumptions. This guide breaks down the confirmed ranking signals based on Amazon's patent filings, official statements, and controlled seller experiments, then shows you how to test what actually moves your products up in search.

How A9 differs from Google's algorithm

A9 optimizes for a fundamentally different goal than Google. Google wants users to find information and stay engaged with search results. Amazon wants users to buy products and complete transactions.

This creates three critical differences:

• Conversion rate weighs heavier than relevance. A product that converts at 18% will often outrank a more "relevant" product converting at 12%, even if the lower-converting product has better keyword optimization.
• Recency matters less. A listing from three years ago with strong sales velocity outranks a new listing with identical optimization but no sales history.
• External backlinks don't help. Unlike Google, A9 ignores off-Amazon authority signals. Your product's New York Times feature does nothing for ranking unless it drives trackable Amazon traffic that converts.

Amazon confirmed these priorities in a 2018 statement: "A9's primary objective is to help customers find and discover products they want to buy. We measure success by how often customers purchase a product after searching."

The three-pillar ranking framework

A9 ranking factors fall into three categories, each with different weight and controllability:

Understanding which pillar to optimize depends on your product's current position. New listings need relevance work first. Established products with stagnant rankings need conversion improvements. Products in competitive categories need authority signals to break through.

Conversion signals: The dominant ranking factor

Conversion rate — the percentage of people who buy after viewing your listing — is the single strongest ranking signal. Amazon's algorithm interprets high conversion as proof that your product matches searcher intent better than competitors.

Unit session percentage (the metric Amazon actually tracks)

Amazon doesn't use traditional conversion rate. Instead, A9 tracks unit session percentage: units ordered divided by sessions (unique visits in a 24-hour period). A customer who buys three units in one session contributes a 300% unit session percentage.

This metric appears in your Business Reports under "Detail Page Sales and Traffic." Benchmarks vary by category, but products ranking in top 10 positions typically show unit session percentages between 12% and 25%.

Factors that drive conversion rate in A9's model

A9 doesn't just track your conversion rate — it tracks the signals that predict conversion, giving those signals direct ranking weight:

• Price competitiveness relative to search results page. If your product is the third-most expensive on page one, A9 factors that into your ranking independent of whether customers actually buy. The algorithm predicts lower conversion and adjusts position accordingly.
• Review count and rating. Products with 500+ reviews and 4.5+ stars get ranking credit even when conversion rates are identical to lower-reviewed competitors. The algorithm treats reviews as a conversion predictor.
• Prime eligibility. FBA products receive a ranking boost separate from their conversion advantage. Non-Prime listings converting at 15% often rank below Prime listings converting at 13% for the same keyword.
• Answered questions. Products with 20+ answered questions in the Q&A section show measurably higher rankings than products with identical conversion rates but fewer answered questions. Amazon's patent US9665881B1 explicitly names Q&A completeness as a ranking input.
• Image count and quality. Listings with 7+ images rank higher than listings with 3-4 images when other factors are equal. A9 uses computer vision to assess image quality — blurry, pixelated, or poorly-lit images receive negative scoring.

Testing framework for conversion improvements

Isolating conversion factors requires controlled testing. Here's the methodology sellers with 50+ weekly sessions can use:

1. Establish baseline. Record your current organic rank for your top 3 keywords and your 14-day average unit session percentage. Use Helium 10 or Jungle Scout rank tracking — manual checks miss intraday fluctuations.
2. Change one variable. Add 3 lifestyle images, or drop price by 8%, or add 15 answered questions. Make the change significant enough to move conversion but change only one thing.
3. Wait 14 days. A9 updates ranking based on trailing conversion data. Changes show impact after roughly two weeks.
4. Measure rank movement. If your average position for "yoga mat non-slip" moved from 18 to 12, and unit session percentage increased from 11% to 14%, you've confirmed that variable affects ranking.
5. Revert and confirm. Return to the original state. If ranking drops back toward baseline, the factor is confirmed.

This process takes 4-6 weeks per variable but produces data you can rely on. Changing five things simultaneously and seeing rank improvement tells you nothing about what worked.

Relevance matching: How A9 determines what your product is about

Relevance signals tell A9 which searches your product should appear in. Unlike conversion factors, relevance is entirely under your control and updates within 24-48 hours of changes.

Keyword placement hierarchy

Not all keyword locations carry equal weight. Based on listing experiments where sellers deliberately placed keywords in different fields and tracked impression changes:

1. Product title — highest weight, especially the first 80 characters
2. Backend search terms — medium-high weight, no difference between positions within the field
3. Bullet points — medium weight, front-loaded bullets weigh slightly more
4. Product description — low weight, A+ Content appears equivalent
5. Product type and category attributes — variable weight, depends on how Amazon has structured that category's browse tree

Two critical rules: A9 ignores keyword frequency. Writing "yoga mat" five times doesn't help. A9 also ignores stop words in phrases — searching "mat for yoga" matches products optimized for "yoga mat."

Backend search terms optimization

The backend search terms field (under "Keywords" in Seller Central) accepts up to 249 bytes. Each character consumes one byte, spaces included. Here's what matters:

• No repetition needed. If "stainless steel" appears in your title, don't repeat it in backend terms.
• Synonyms and alternate spellings work. "Yoga mat" in the title + "exercise mat" in backend = you rank for both.
• No commas or punctuation. A9 parses backend terms as a single string and generates all possible combinations. Writing "water bottle insulated stainless" makes you eligible for "insulated water bottle," "stainless water bottle," "insulated stainless bottle," etc.
• Misspellings don't help. Amazon's search system automatically corrects common misspellings. Adding "excercise mat" wastes space — searches for "excercise" already surface products optimized for "exercise."

The category browse node impact

Your product's category assignment affects which searches you're eligible for. Miscategorized products rank poorly even with perfect keyword optimization.

Example: A seller listed a yoga mat in "Sports & Outdoors > Exercise & Fitness > Yoga > Mats." Moving it to "Sports & Outdoors > Exercise & Fitness > Exercise Mats" — a broader category — increased impressions for generic "exercise mat" searches by 40% within three days. Category structure varies by marketplace, so test your specific browse path.

Relevance testing framework

Testing relevance changes is faster than testing conversion changes:

1. Identify target keyword. Pick a keyword with 1000+ monthly searches where you currently don't rank in the top 100.
2. Add keyword to title or backend terms. If adding to title, place it in the first 80 characters.
3. Wait 48 hours. Check if you now appear in search results for that keyword. Use an incognito browser or rank tracker.
4. Monitor impression changes. In Business Reports, compare 7-day impression counts before and after the change for that specific keyword (if you run auto campaigns, the Search Term Report shows keyword-level impressions).

If impressions increase but conversion rate drops, the keyword may be attracting irrelevant traffic. A common mistake: optimizing for high-volume keywords that don't match your product's actual use case.

Seller authority: The overlooked ranking multiplier

Seller-level metrics create a ranking multiplier applied to individual listings. Two identical products with identical conversion rates and keyword optimization rank differently if one seller has stronger authority signals.

Account health metrics that affect ranking

Amazon's Seller Performance dashboard tracks these metrics. Each impacts ranking to varying degrees:

• Order Defect Rate (ODR) below 1% — required for normal ranking treatment. Accounts with ODR above 1% see suppressed rankings even when individual listing metrics are strong.
• Late shipment rate below 4% — FBM sellers only. Late shipments signal poor reliability, reducing ranking across all listings.
• Pre-fulfillment cancel rate below 2.5% — canceling orders due to inventory errors signals poor operations. Chronic offenders see ranking penalties.
• Valid tracking rate above 95% — FBM sellers must upload tracking for 95%+ of orders. Missing tracking prevents listings from ranking competitively.

These thresholds appear in Amazon's Fair Pricing Policy and Seller Performance Standards documentation. Violations don't just risk suspension — they suppress rankings immediately.

Brand Registry benefits for ranking

Brand-registered sellers get measurable ranking advantages:

• A+ Content appears to improve conversion rate prediction. Branded listings with A+ Content rank higher than non-branded listings with identical actual conversion rates, suggesting A9 treats A+ Content as a positive signal independent of conversion impact.
• Sponsored Brand ads influence organic rank. Running Sponsored Brand campaigns for 30+ days correlates with organic rank improvements for those keywords, even when ad spend stops. The mechanism likely involves A9 interpreting ad engagement as a relevance signal.
• Brand analytics data access. Brand Registry provides search query performance data that non-registered sellers can't access, enabling better keyword optimization.

Sales velocity and ranking momentum

A9 weighs recent sales velocity more heavily than historical performance. A product selling 30 units daily for the past week outranks a product that sold 40 units daily last month but only 15 units daily this week.

This creates a momentum effect: products ranking well continue ranking well because high visibility drives more sales, which maintains ranking. The inverse is also true — a ranking drop reduces visibility, which decreases sales, which further drops ranking.

Breaking this cycle requires external traffic. Launching PPC campaigns, influencer promotions, or deal events can inject enough sales velocity to push organic rank upward, after which organic traffic sustains the new position.

Factors that don't matter (despite common claims)

Several supposed ranking factors circulate in Amazon seller communities without evidence:

• Time of day for listing updates. Some sellers claim updating listings at 2 AM Pacific helps ranking. A9 processes listing changes continuously — timing makes no difference.
• Character count in bullet points. As long as bullets aren't truncated (under 1000 characters per bullet on desktop), length doesn't affect ranking. Shorter, scannable bullets often improve conversion, which indirectly helps ranking.
• Seller feedback count. Individual seller feedback (the seller rating, not product reviews) shows no correlation with product ranking in controlled tests. Account health metrics matter, but feedback count does not.
• Number of variations. Having 8 color variations versus 3 doesn't improve ranking unless the additional variations increase total sales velocity.
• Video count. Listings with 5 videos don't rank higher than listings with 1 video when conversion rates are equal. Videos help conversion, which indirectly helps ranking, but video count itself isn't a ranking input.

How PPC campaigns interact with organic ranking

Sponsored Product campaigns affect organic ranking indirectly through three mechanisms:

1. Sales velocity injection. PPC sales count toward total sales volume, which A9 uses as a ranking signal. Products with combined organic + PPC velocity of 50 units daily rank higher than products with 30 organic-only units daily.
2. Conversion rate data. PPC campaigns generate conversion data for keywords you don't organically rank for yet. High PPC conversion rates for a keyword signal to A9 that your product is relevant, improving organic eligibility for that term.
3. Review accumulation. PPC-driven sales generate reviews faster, which improves conversion rate predictions, which improves organic rank.

The inverse isn't true: organic ranking doesn't directly improve PPC performance. Ad position depends on bid amount and ad quality score (a separate algorithm from A9's organic ranking system).

Testing PPC's organic impact

Isolating PPC's effect on organic ranking requires this protocol:

1. Establish baseline organic rank. Track your position for 5 target keywords where you currently rank between positions 20-50.
2. Launch exact-match PPC campaigns for those keywords. Run campaigns for 30 days with daily budgets sufficient to generate 10+ clicks per day per keyword.
3. Monitor organic position weekly. Use rank tracking software to record position changes throughout the campaign period.
4. Pause PPC and observe decay. After 30 days, pause campaigns and track whether organic rank holds, improves, or declines over the next 14 days.

Products that maintain improved organic rank after PPC pause have successfully built ranking momentum. Products that drop back to baseline positions were likely ranking due to PPC-driven conversion rate improvements that didn't translate to organic clicks.

Category-specific ranking variations

A9's weighting of ranking factors varies by product category. Amazon adjusts the algorithm based on customer behavior patterns in each category.

Consumables and replenishment products

Categories like Health & Household, Grocery, and Pet Supplies show higher ranking weight on Subscribe & Save enrollment rates. Products with 20%+ subscription rates rank higher than products with identical conversion rates but lower subscription adoption.

These categories also show faster ranking decay. A supplement that stops selling for two weeks drops ranking more severely than a kitchen gadget with the same sales interruption.

Fashion and size-variant categories

Clothing, Shoes, and Jewelry show higher ranking weight on return rates. Products with return rates above category averages receive ranking penalties even when conversion rates are strong.

Size-variant listings also benefit from completeness — having stock across 8 sizes ranks better than having stock in only 3 sizes, independent of total unit velocity.

Electronics and technical products

Categories like Electronics, Computers, and Camera & Photo show higher ranking weight on Q&A section completeness. Products with 50+ answered questions rank measurably higher than products with fewer than 10 questions when other factors are equal.

These categories also see stronger ranking correlation with A+ Content usage, likely because technical specifications communicated in A+ Content improve conversion rates.

Monitoring ranking changes and diagnosing drops

Ranking isn't static. Weekly monitoring catches problems before they compound.

Tools for rank tracking

Manual rank checks miss the full picture. Use automated tools that check your position multiple times daily:

• Helium 10 Keyword Tracker — tracks up to 2500 keywords across multiple products, shows position changes by date
• Jungle Scout Rank Tracker — monitors rank + estimates organic search volume for each keyword
• Seller Central Brand Analytics (Brand Registry only) — shows search frequency rank for top customer search terms finding your products

Common ranking drop causes and fixes

Recovery timeline expectations

Ranking recovery isn't immediate. After fixing the root cause:

• Relevance fixes (keyword additions) — 24-48 hours to regain impressions
• Conversion fixes (images, price changes) — 7-14 days to see rank improvement
• Stockout recovery — 14-30 days to return to pre-stockout rank, depending on stockout duration
• Account health recovery — 30-60 days after metrics return to good standing

Ranking doesn't snap back to previous positions. A9 requires sustained proof that the fixed listing deserves better placement.

Testing your own ranking factors

Amazon doesn't publish A9's exact algorithm, and it evolves continuously. The most reliable ranking insights come from your own controlled tests on your own products.

Minimum requirements for valid testing

Not every product is suitable for ranking experiments:

• Minimum 50 sessions per week — lower traffic makes conversion changes too noisy to interpret
• Stable inventory — stockouts during test periods invalidate results
• No major external campaigns — don't test during Lightning Deals or influencer launches that skew traffic patterns
• Single-variable changes only — change one thing, measure, revert, confirm

Building your ranking factor database

Systematic testing over 6-12 months builds institutional knowledge about what works for your specific products:

1. Document baseline data. Record rank, conversion rate, sessions, and unit session percentage before each test.
2. Track correlation, not just rank. If rank improves but conversion rate drops, the change attracted wrong-intent traffic.
3. Test seasonal effects. A ranking change in November may not replicate in March due to demand shifts.
4. Share findings across your catalog. A successful image style test on Product A likely applies to Products B and C in the same category.

Sellers who treat ranking optimization as hypothesis-driven experimentation outperform sellers who chase generic best practices.

What to optimize first

Different products need different optimization priorities. Here's the decision framework:

If your product is new (under 30 days old, fewer than 15 reviews): Focus on relevance. Fix title and backend keywords first, then drive external traffic through PPC to build initial sales velocity. Conversion optimization can't help until you're getting enough traffic to measure.

If your product ranks between positions 15-40 for target keywords: Focus on conversion rate. You're already relevant enough to show up in searches — now you need to prove you convert better than products ranking above you. Test images, pricing, and review count.

If your product ranks in top 10 but stagnates: Focus on seller authority and competitive defense. Enroll Brand Registry if you haven't, launch Sponsored Brand campaigns, and monitor competitor changes weekly. At the top of page one, small conversion improvements from competitors can displace you.

If your product dropped suddenly: Fix operational issues first. Check for stockouts, listing suppressions, or account health problems before testing optimization changes. Ranking drops usually signal problems, not opportunities.