Amazon Sales Rank, Secret Algorithm Exposed

amazon sales rankAmazon Sales Rank? A lot of Indie authors want to know how it’s determined. For instance, how many ebook sales does it take to reach the 10,000 mark, or the 1,000 mark or *gasp* the top 100 best sellers? Like Fermat’s Last Theorem and Coke’s formula, Amazon Sales Rank algorithm seems to be a highly guarded secret.

Unfortunately, Amazon won’t give us an X + Y + Z = Sales Rank equation. They have said, “As an added service for customers, authors, publishers, artists, labels, and studios, we show how items in our catalog are selling. The lower the number, the higher the sales for that particular item. The calculation is based on Amazon.com sales and is updated each hour to reflect recent and historical sales of every item sold on Amazon.com.”

The first question: how many ebooks does Amazon sell? There are plenty with no sales ever, therefore no sales ranking, which makes it difficult to know how many are really out there. 5 million ebooks? 7 million? 10 million? Since books have shown up in the 4 million range with at least one sale, it’s safe to assume this number is reaching 10 million.

Save 92% Sell Ebooks on Amazon and Major Retailers

The following appear to be true (estimates based on Amazon US Kindle ebooks);

  • a book must have a sale (even just one) to be ranked.
  • a few sales will leap a ranking from the millions to the hundred thousands, as in going from 4,156,977 to 357,954 after just 3 sales.
  • recent sales count much more than past sales.
  • to break the 10,000 or 1,000 or the top 100 marks – the sales have to multiply in geometric progression (a heck of a lot).
  • to be in the top 100 best selling books, sales must routinely be well over 200 sales/day (estimating here).
  • sales updated from the recent hour appear to be the most important factor.
  • price doesn’t appear to matter, only total number of sales do.

math equationsOkay, this can’t be rocket science. Sales happen over time. Factors are pretty limited: total number of sales ever (history) and recent sales (the past month, week, day, hour). Since Amazon updates rankings hourly, it can be deduced that Amazon places at least 2 major elements into the equation: the total number of sales ever, history, and the number of sales hourly. In my experience of obsessing overs the ranking as it moves up and down, there have been many hours where some of my books didn’t sell any copies and the ranking only changed slightly, so it also seems logical that daily sales and weekly or even monthly sales must play into the equation. All of these factors must play a role, some more than other and perhaps from hourly to daily to weekly to monthly in importance. That’s got to be close; what else could possibly go into the equation?

Here’s a telling comparison. The #1 best-selling book from 2009 (New Moon, Book 2 of Twilight Series by Stephanie Meyer) shows a current sales rank of #936. Another book that was published just 4 months ago in May 2011 (The Mill River Recluse by Darcie Chan) is currently ranked #2 on Amazon. Yet another book published last month in August 2011 (Thunder Dog by Michael Hingson) is ranked #4. There is no way the latter books have anywhere near the total sales as New Moon. This indicates that hourly sales are far more important to the equation than total sales ever (history). Apparently Amazon prioritizes what is hot right now and makes hourly sales the main factor. In fact, 4 of the top 10 bestsellers were published this year while the oldest book in the top 10 is from 2002, another is from 2007 and the rest from 2009 or more recent. Ah-hah, time is by far the prime ingredient. (Great news for Indie authors; get your sales rolling and you could make the top of the charts as the evidence shows.)

Below could be the secret equation to determine sales ranking:

hourly (# of sales) x daily/2 x weekly/3 x monthly/4 x history/10 or more = Sales Corresponding Coefficient (SCC)

This way the main ingredient is time, namely hourly and daily, with slightly less emphasis on weekly and even less on monthly and last on history. The SCC of any book with substantial sales can be a huge number; the highest number would have the lowest sales ranking. It should also be noted than a zero (0) for any of these categories other than history should not make the formula reduce itself to zero. Perhaps every other category besides history should have a 1 added to the equation to force a real number to be the outcome.

What happens to books with the exact same number of SCC? Whichever book has been on the shelf longer would rank lower.

What about price affecting ranking? This has been mentioned but seems seriously doubtful. Evidence does suggest that lowering the book’s price can lead to more sales and free books get downloaded like crazy, but it’s hard to imagine Amazon would want to encourage the cheaper books to have a better ranking as they would make less money that way. Could the opposite be true, could higher priced books get better ranking? Perhaps but doubtful again, since the top 10 bestsellers currently have 4 books priced at 99 cents.

This theory is my opinion: Amazon sales ranking places priority over recentness of sales by hour, day, week and month while also factoring in a small percentage for total sales history. It’s a a fun thing to conjecture upon but probably not too helpful in selling ebooks. If you want to read more on the subject by some pretty focused people who have spent far more history on this, a Google search or these links might be useful – http://ebmv.blogspot.com/ and http://futureperfectpublishing.com/2008/01/27/dark-mysteries-of-the-amazon-sales-rank/.

Add a comment?


HomeVideosHow-To GuidesNovelsAbout


Subscribe to this blog for updates on indie authors and self-publishing.

add me to Google Plus circles

Advertisements

18 Responses to “Amazon Sales Rank, Secret Algorithm Exposed”

  1. Stephanie Says:

    Fascinating Jason! The price affecting ranking is certainly something to think about.

  2. Did Life Just Get Harder for the Indie Author? (Ideas on How to Cope) | Self-Published Author's Lounge Says:

    […] I am not an expert on Amazon sales rankings.  I did, however, find this article on the subject: https://ebooksuccess4free.wordpress.com/2011/09/13/amazon-sales-rank-secret-algorithm-exposed/.  If you aren’t following the “How to Make, Market and Sell Ebooks – All for […]

  3. Borrowind Says:

    The Amazon rankings just don’t make any sense to me. One best selling book, shifting about 30 copies a month with a healthy mark-up, languishes at a Rank of 120,000. Yet another worst selling book, selling only 2 copies a month at 99 cents, is at a Rank of 82,000. Go figure. The other problem is that people often simply can’t find things on the Kindle Store. Only the direct link gets to that bestseller book’s page. No combination of keywords on earth will get anyone to it via the Amazon search box in a browser. But it _does_ appear if they search from the Kindle itself. Again, go figure… Amazon’s store search and rankings calculations are obviously profoundly broken.

    • Jason Matthews Says:

      Thanks for the comments. It would help me to see your ranking points with the actual titles you’re referring to. Are these your books or are you getting sales figures from a reliable source? Some sales sources are inaccurate, I have found. Also, which keywords and title are you searching for? For example, I have found that typing in the search phrases, “sell ebooks” or “make free website” will bring two of my books to the top of the results page, but I admit it helps that those words are within the title of the books.

  4. Lucas Nicolato Says:

    I’m a software engineer who never worket at amazon, but if were the one responsible for figuring out that algorithm I could probably end up with something like that.

    Start with a “value”, say X, of 1.
    Every time a book is sold, add X to a “Total Sales” variable.
    Every hour, multiply the current X by a small incresing factor, say 1.001
    Now, everytime you want a sales rank, just query your database ordering by “Total Sales”.

    Last hour’s sales will end up having a far higher weight than last year’s ones, but the entire sales history stills play a part in the rank.

    The behavior of amazon’s sales ranks seems consistent with an algorithm like the above.

    It can be not as much rocket science as it seems at first.

  5. Trappy Says:

    This is TOTALY wrong!!! If the number of sales in any part of the equation is null, then the Sales Coefficient is allways null!
    Thiw equation needs serious revising!

    • Jason Matthews Says:

      Are you mentioning Amazon’s equation that came from their press release? The monthly royalty payment for each KDP Select book is based on that book’s share of the total number of borrows of all participating KDP books in the Kindle Owners’ Lending Library. For example, if total borrows of all participating KDP Select books are 100,000 in December and an author’s book was borrowed 1,500 times, they will earn $7,500 in additional royalties from KDP Select in December. Amazon expects the fund to be at least $6 million for all of 2012, in addition to the $500,000 allocated for December 2011.

      • Trappy Says:

        No, i am referring to the equation mentioned in this post.

        “hourly (# of sales) x daily/2 x weekly/3 x monthly/4 x history/10 or more = Sales Corresponding Coefficient (SCC)”

        • Jason Matthews Says:

          Trappy, that’s correct and the reason why the next line of the article reads with this; “It should also be noted than a zero (0) for any of these categories other than history should not make the formula reduce itself to zero. Perhaps every other category besides history should have a 1 added to the equation to force a real number to be the outcome.”

  6. Trappy Says:

    Jason, you should be taking into account the price of the item. E.g. a book sold 100 times in 1 hour at 15$ is better selling than a book sold 100 times in 1 hour at 0,99$ assuming the more expensive the item sold the more difficult the sale becomes.

    They also might be taking into account the selling price history of each item during certain periods of time…!!!

    • Jason Matthews Says:

      Price has nothing to do with Amazon ranking, I’m pretty sure. Mill River Recluse by Darcie Chan is still ranked #10 and was #4 when I wrote this blog post and only sells for 99 cents.

  7. Trappy Says:

    Price is definitely in the equation of Amazons rating algorithm, its weight is insignificant (in order not to change the actual rank but to distinguish between 2 products with the same rank but of different price). Thats why you arent able to actually see it effects on the rank(Mill River Recluse by Darcie Chan), but the price is there, for sure!

  8. Trappy Says:

    Jasson, i do not have any data to prove it, appart from my experience. If it is not the price that deterines the ranking of 2 products sold equal times over a certain preiod of time, then what is? Antiquity? – I dont see the point in that!

    There are only 3 factors that can be included in this equation
    1. No of Sales
    2. Time
    3. Price

    • Jason Matthews Says:

      Talking about factoring price into the equation to separate two books with exactly the same sales is really a non factor. I seriously doubt it matters because if you consider two books that are both priced the same and both sold just one copy, whichever sold more recently will have a higher sales rank. So in your example, the two books would have to sell not only the same number of copies but those copies would also have to be sold at the same hours.

      • Trappy Says:

        What I am saying is that if two books sell the same number of copies at the same hours then you have to rank them in relation to the price!(I prioritize recency of sale before selling price, i.e. i give it more weight in relation to price) If the price is also the same then you have to add another factor (e.g. Better reviewed) if the review factor is also(!) the same you must add yet another factor (e.g. number of reviews) etc.
        There is always the possibility that 2 items can have the same ranking value, and in order to avoid overlap you must add as many factors as possible to eliminate/minimize the possibility.

  9. Publishing My Book With Amazon and KDP Select: Month 1 - The Tech Savvy Writer Says:

    […] algorithm’s net effect on sales? I’m not sure, though many people have tried to guess. One thing I do know–my book’s trip up and down the ranking […]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: