Amazon sells more books than any other retailer in the world. Most of those sales are not driven by readers searching for a specific title. They are driven by the algorithm deciding which books to show to which shoppers, in which categories, at which moments. If you publish on Amazon and the algorithm does not like your book, almost nothing else you do will matter.
This guide walks through how the Amazon book algorithm actually works, what signals it watches, what authors can do about it, and where the limits are.
What the Amazon Algorithm Actually Does
The Amazon algorithm is not one system. It is several systems working together, each making different decisions about your book.
The search ranking system decides which books appear when a shopper types a query. The recommendation system decides which books to show on product pages, in “customers also bought” carousels, and in personalized homepage feeds. The advertising system decides which paid ads to show to which users. The browse system decides which books appear when shoppers click into a category or subcategory.
Each of these systems uses similar signals, but they weight them differently. A book can rank well in search and poorly in recommendations, or vice versa. Knowing which system is driving your traffic tells you what to fix when sales drop.
The algorithm Amazon used in its early years (sometimes called A9) has been heavily modified over the past several years. The current generation, often called A10 in industry shorthand, gives more weight to external traffic, conversion patterns, and reader behavior than the older system did. Amazon does not publish exact rules, so most of what authors know comes from observed patterns rather than confirmed documentation.
Signals the Algorithm Watches
Several factors consistently move books up or down in Amazon’s systems.
Sales Velocity
Sales velocity is how many copies your book sells in a given window, often the last twenty-four hours, the last seven days, and the last thirty days. The algorithm uses velocity to gauge demand and adjust where it shows your book.
Velocity matters more than total sales. A book selling ten copies a day consistently for a month outranks a book that sold three hundred in one day and nothing since. The algorithm rewards steady demand over spikes.
This is why launch strategies that concentrate sales into a tight window pay off. A book selling fifty copies a day for three days in launch week climbs the rankings faster than the same fifty copies spread over a month.
Conversion Rate
When a shopper lands on your product page, do they buy. The conversion rate, meaning how many of your page visitors purchase, tells Amazon how well the book matches the search or browse intent.
Books with strong conversion rates get shown more often to similar shoppers. Books with weak conversion rates get demoted, even if they sell well on absolute volume.
Conversion is driven by the cover, the title, the price, the description, the reviews, and the sample. The algorithm watches all of these by watching shopper behavior on the page.
Reviews & Ratings
Average rating and review count both feed into algorithmic decisions. The exact weighting is debated, but the pattern is clear. Books at 4.3 stars and above with twenty-plus reviews see more algorithmic favor than books at 3.7 stars with five reviews.
What Amazon watches in reviews has shifted. It used to be just the star rating. Now the system looks at review velocity (new reviews coming in steadily), review text patterns (specific reader signals like genre fit), and verified purchase ratio.
Click-Through Rate
When your book shows up in search results or browse pages, do shoppers click. The click-through rate is part of how the algorithm decides if to keep showing your book in those positions.
The cover does most of the work here. A cover that does not stand out in its category gets passed over, and the algorithm learns to show it less often.
Read-Through & Engagement
For books in Kindle Unlimited, the algorithm watches how much of the book readers actually finish. Books with high read-through rates earn more page-read royalties and rank better. Books that readers abandon after twenty pages get demoted.
This signal does not exist for non-KU books, since Amazon cannot see how much of your book paid customers actually read.
External Traffic
Amazon now favors books that bring outside traffic to the platform. A book whose visitors come from a podcast appearance, a newsletter feature, or a viral social post signals to the algorithm that the book has an active audience driving demand.
This is one of the biggest changes from the older algorithm generation. External traffic was not weighted heavily in the past. Now it is one of the strongest signals.
What Authors Can Actually Influence
Some signals are within author control. Others are not.
Things You Can Move
Your cover, title, subtitle, and description directly affect conversion rate. Strong copy and design push these numbers up. Weak versions hold them down.
Your categories and keywords decide where the algorithm even gets a chance to show your book. Picking wrong here is one of the most common mistakes.
Your price affects conversion. Books priced above genre norms convert worse. Below norms can signal low quality.
Your reviews depend partly on the book itself and partly on your launch strategy. Authors who systematically ask early readers for reviews see higher counts than those who hope reviews show up.
Your external traffic depends entirely on you. Email lists, podcast pitches, social posts, and partnerships all bring outside readers to your Amazon page.
Things You Cannot Move
The algorithm itself. Amazon changes its rules without warning. A category that worked last year may not this year. A keyword strategy that drove sales in 2024 may underperform now.
Competing books. If a giant in your category launches a new release, your sales may dip regardless of what you do.
Reader behavior. You cannot make people read your book to the end, even if it deserves it.
Working with the algorithm means focusing on the signals you can influence and not stressing about the ones you cannot.
Launch Strategy and the Algorithm
The pattern of your launch teaches the algorithm what to expect from your book.
A weak launch with slow sales, few reviews, and no external traffic signals that the book has low demand. The algorithm shows it to fewer shoppers, which produces fewer sales, which reinforces the signal. Books that launch this way often never recover.
A strong launch with concentrated sales, fast reviews, and external buzz signals high demand. The algorithm shows the book to more shoppers, which produces more sales, which reinforces the signal. Books that launch this way often keep selling for months.
The first thirty days set the trajectory for the year. The first ninety days set the trajectory for the life of the book. Most books that earn substantial money make most of that money in the first six to twelve months, and the launch sets the ceiling.
How Categories Affect Algorithmic Visibility
Categories are the rails the algorithm runs on. When you choose your categories, you tell Amazon which subset of shoppers should see your book and which competitor books you sit alongside.
Picking categories that are too broad puts you in pools where you cannot rank. Picking categories that are too narrow puts you in front of shoppers who are not in the market for your kind of book.
The sweet spot is a category small enough to rank in the top hundred with realistic sales, but large enough that the shoppers in it are people who actually want your book. Most working authors aim for categories where reaching the top hundred is achievable with twenty to fifty sales per week.
Amazon lets you select multiple categories for each book through the KDP support channel. Using all available slots gives the algorithm more chances to surface your book in different contexts.
How Keywords Feed the Algorithm
Keywords are the search terms the algorithm pairs your book with. You enter up to seven keyword slots in KDP, and Amazon uses them to decide which searches should include your book.
Working keyword strategy matches the phrases real shoppers use. Generic terms like “fantasy novel” or “self-help book” pit you against millions of competitors. Specific phrases like “found family fantasy with magic” or “self-help for entrepreneurs over forty” put you in front of shoppers who already know what they want.
The algorithm also pulls keywords from your title, subtitle, description, and reviews. Keywords appearing across all these places signal strong relevance to a search term.
Why Reviews Move the Algorithm
Reviews influence the algorithm in three ways.
Star rating affects if shoppers click. A 4.5-star book gets clicked more often than a 3.5-star book in the same search results, all else equal.
Review count creates social proof that affects conversion. Shoppers see twenty reviews and assume the book has been read by enough people to be safe to buy. Shoppers see two reviews and hesitate.
Recency tells the algorithm the book is still active. A book that gets reviews steadily over months looks alive. A book with all its reviews from launch week and nothing since looks stale.
Building reviews systematically (through advanced reader teams, follow-up emails to verified buyers, and prompts at the back of the book) keeps this signal active.
How External Traffic Changes the Math
The shift toward favoring external traffic has changed the game for working authors.
In the old algorithm, building an audience on Amazon meant working within Amazon. Ads brought traffic. Free promotions brought traffic. Cross-promotion with other Amazon authors brought traffic. Amazon was the system, and the loop stayed inside.
In the current algorithm, outside traffic now signals demand more reliably than internal traffic does. A reader who comes to your Amazon page from a podcast is treated as a stronger signal than a reader who clicked an Amazon ad to get there.
This means authors with email lists, podcast appearances, newsletter coverage, and content followings have a structural advantage. They bring traffic that Amazon’s system reads as organic demand, and the algorithm responds by giving the book more visibility to shoppers already on Amazon.
For authors without an outside audience, the workaround is to build one. Newsletter swaps with other authors in your genre, guest podcast appearances, and content publishing on platforms outside Amazon all feed the same signal.
When the Algorithm Stops Working for Your Book
Books that sold well sometimes stop. The algorithm has decided to show them less, and the question is why.
Common reasons include the category getting crowded as bigger releases entered your space and took the visibility you had. Or the reviews stopped coming in, and the algorithm reads silence as fading demand. Or the cover and description got stale, conversion dropped, and the algorithm responded by demoting the book.
Sometimes you stopped sending external traffic. The algorithm noticed and adjusted. Sometimes a platform change happened. Amazon updates its system regularly, and the rules shift in ways that hurt your book.
The fix is usually a combination of refreshing the listing, building new reviews, and restarting external traffic flow. Books that stalled can come back, but only with active work.
Working With the Algorithm, Not Against It
The authors who do well on Amazon long-term treat the algorithm as a partner they have to keep informed. They send strong signals consistently: steady sales, fresh reviews, external traffic, optimized listings, and engaged readers.
The authors who struggle treat the algorithm as a mystery to crack or a system to game. Neither approach works. The algorithm rewards books that look like real demand for real books, and the best way to look like that is to be that.
For authors starting out, the move is to focus on the basics. Pick the right categories. Write strong copy. Get the cover right. Build reviews systematically. Drive outside traffic. Repeat. The algorithm will do its part if you do yours.
