How Amazon's Algorithm Actually Works (And What Authors Can Do About It)

Most authors think of Amazon as a bookstore. It is better understood as a search engine that happens to sell books. Like Google, it has an algorithm that decides which books to show readers, in which order, and how prominently. The algorithm is not random. It responds to specific signals, and authors who understand those signals can influence their results in concrete ways.

Sales Velocity Is the Core Signal

The single most important factor in Amazon's ranking system is sales velocity: how many copies of a book are sold in a short window of time. A book that sells 100 copies in one day will rank significantly higher than a book that sells 100 copies over one month. Amazon interprets rapid sales as evidence of demand, and it rewards that with more visibility.

This is why launch strategy matters so much. A coordinated launch week with a full email blast, Amazon Ads, and coordinated ARC reader reviews can create a spike in sales velocity that pushes a book into prominent positions in its categories. Without that coordinated push, the launch window passes without building any algorithmic momentum.

Keywords and Categories

Amazon allows authors to enter 7 keyword fields in their KDP dashboard. These are not tags or hashtags. They are matching signals that determine which reader searches surface your book. The most effective keywords are 3 to 5 word phrases that reflect how readers actually search. Phrases like "dark romance enemies to lovers" or "psychological thriller domestic suspense" are far more effective than single words like "romance" or "thriller."

Categories work differently. Amazon allows 2 primary categories chosen at upload, but authors can request up to 10 additional categories through KDP support. Choosing narrower subcategories dramatically increases the chance of reaching the Best Seller badge, which then appears as visible social proof on the product listing. A book that reaches number one in a subcategory with moderate competition can carry the "Amazon Bestseller" label in its marketing materials permanently.

Reviews as Both Trust and Algorithm Signal

Amazon's recommendation engine begins actively promoting a book once it crosses roughly 15 to 25 reviews with a strong average rating. Below that threshold, most recommendation features are essentially switched off. The 50-review mark is where the algorithm begins placing a book in "Customers Also Bought" carousels alongside comparable titles.

Review velocity matters as much as total count. Ten reviews posted in launch week carry far more algorithmic weight than ten reviews posted over six months. This is why ARC reader programs, where advance copies go out to readers who commit to reviewing near launch, are such a critical part of the launch strategy.

Conversion Rate and Click-Through

Amazon tracks what percentage of readers who see a book actually buy it. A product page with a weak cover, a vague description, and few reviews will have a low conversion rate. Amazon interprets low conversion as low relevance and stops showing the book as frequently. This means a poor product page actively harms discoverability over time.

The reverse is also true. A product page that converts well gets shown more often. This is why investing in a genre-appropriate cover, a properly structured book description, A+ Content, and editorial reviews is not cosmetic. It is algorithmic infrastructure that determines long-term visibility.

Also-Bought Clusters and Recommendation Positioning

One of the most valuable things Amazon can do for a book is place it in the "Customers Also Bought" section of comparable bestselling titles. When readers who bought a popular thriller in your genre also buy your book, Amazon's algorithm connects the two and starts recommending your book to other readers who purchased the comparable title.

Authors can influence this directly through Amazon's ASIN targeting feature in their ad campaigns. By running ads that appear on the product pages of comparable bestselling books, authors increase the chance that the same readers who are interested in those titles also discover and purchase their book, building the also-bought association over time.

Kindle Unlimited and KDP Select

Enrolling a book in KDP Select places it in Kindle Unlimited, where subscribers can read it for free and the author earns per page read. The payout rate is roughly $0.004 to $0.005 per page, which means a fully read 300-page book generates about $1.20 to $1.50. The tradeoff is exclusivity. An ebook in KDP Select cannot be sold on Apple Books, Kobo, Barnes and Noble, or any other platform for 90-day rolling periods.

KDP Select works best for genre fiction authors with series, particularly in romance, thriller, and fantasy, where Kindle Unlimited has an active readership. It is less well-suited for nonfiction, literary fiction, or authors with an existing audience across multiple retail platforms.