Amazon’s sales rank algorithm is surprisingly simple…
Original article written by John Doppler of Selfpublishingadvice.org – link provided below.
1. Each sale or download of a product counts as one point toward a hypothetical “rank score”.
2. Each day, the preceding day’s score decreases by half, and is added to today’s points.
3. For each category on Amazon, books are ranked based on their current scores.
Monday, a book sells 32 copies. That’s 32 points towards its ranking.
Tuesday, the book sells 36 copies. Those 36 points are added to half of Monday’s total (32 / 2 = 16 points), for a total of 52 points.
Wednesday, the book sells 16 copies. Those 16 points are added to half of Tuesday’s total (52 / 2 = 26 points), for a total of 42 points.
…but the devil is in the details.
The underlying process is simple, but there are aspects of the process that contribute to the “strange” behavior of the algorithm.
Sales rank is relative to other books.
A book does not exist in a vacuum. As your book rises in sales rank, it will displace other books. As other books rise through the ranks, your book may be pushed downwards. The constant churn in placement leads to the phenomenon where a book that hasn’t sold any copies rises in sales rank: the score of one or more books above it have declined at a faster rate, pushing the other books upwards in the rankings.
This counterintuitive feature of the algorithm is responsible for more confusion than any other.
The more recent the sale, the more weight it has.
Because the formula weights sales by recency, the effect of a sales spike quickly fades. The algorithm favors steady sales over a dramatic surge.
Consider the two books below (Figure 1). Book A experiences slow, constant growth for the first two weeks. Book B offers a promotion which results in an explosion of sales, but those sales quickly settle back to normal levels once the promotion ends.
Reviews, ratings, and price do not affect sales rank.
Sales rank is governed by sales and downloads, with a little adjustment by Amazon’s algorithms. Sales rank is unaffected by the number of reviews, ratings, or any aspect of a book other than its sales performance.
Enrollment in KDP Select/Kindle Unlimited does not confer any direct advantage to sales rank.
Titles in KDP Select do not receive higher placement just for enrolling in the program. However, downloads of books through Kindle Unlimited and Kindle Online Lending Library are treated as sales, and they are credited immediately.
A download is immediately recorded as a sale in the sales rank algorithm, regardless of what percentage of the book is read.
High sales rank does not guarantee high placement in search results.
Sales rank plays a very minor role in determining the order of Amazon search results. Other factors such as relevance, keywords, sales history, product listing quality, and available inventory may influence Amazon’s algorithms. Therefore, a book with high sales rank may appear later in search results than lower-ranked books.
It takes twice as many sales to hit a rank than it does to maintain it.
Because each day’s sales rank builds on previous sales, twice as many sales are needed to initially achieve a rank as compared to maintaining that rank. In other words, an author may need 40 sales to initially hit a given rank, but once that momentum is established, they will only need 20 sales per day to maintain that score. (Sales rank may still fluctuate due to the performance of other titles around it.)
Similarly, if no sales are made, the book’s score is cut in half on the following day.
Twice as many sales are needed to hit a rank initially; each day without sales halves a book’s score.
Pre-orders are counted immediately.
Pre-orders are counted on the day the book is ordered, rather than on the date of the book’s release. This explains how books that have not yet been released may have a high sales rank, a common source of confusion.
Sales momentum is a key factor in the algorithm. The early boost from pre-orders have the potential to propel a title onto the Amazon charts faster and for a longer period of time than a launch day blitz alone would.
Original article written by John Doppler of Selfpublishingadvice.org and the entire article can be found here.