Ever wonder how Amazon’s ranking algorithms work? Why one book gets recommended to readers and another doesn’t? The difference between the popularity lists and the bestseller lists? Well, indie author Edward W. Robertson doesn’t work for Amazon, but he’s a stats junkie who’s been studying the e-seller’s algorithms for a while. Today, he’s here to answer questions on how books get ranked and recommended along with new changes that could punish 99-cent titles and take some of the appeal out of the KDP Select program.
Amazon Algorithms Examined
You seem to enjoy studying how Amazon’s algorithms work for ranking and recommending books. Before we talk about what’s new, could you give authors an idea of how things work, at least insofar as you know? What goes into ranking a book and causing it to appear on such-and-such bestseller list? Also, what’s the difference between the bestseller lists and the popularity lists?
Yeah, studying these things is a lot of fun for me. I don’t have any formal training in numbermancy (which I’m pretty sure is the term), but spending the last 10-12 years reading about the statistical study of baseball appears to have taught me a few things about data analysis. Perhaps I wasn’t wasting my life after all!
On to the lists. Everyone who’s spent much time on the Kindle store has seen both the bestseller and the popularity lists. The bestseller list is the Top 100 of a given category of books. For instance, here’s the bestseller list for Epic Fantasy.
The popularity list is the list of all books in that category. Here’s the popularity list for Epic Fantasy.
What you’re currently seeing on those lists will depend on when you’re reading this, but you’ll note they aren’t identical. That’s because they differ in key ways. The bestseller list is essentially a gauge of how many copies a book has sold over the last 24 hours. It takes longer-term sales into account to a degree, but the last 24 hours are far and away the most important factor. A book can rise and fall extremely swiftly on the bestseller list.
The popularity list is more complicated. For one thing, Amazon changes the formula for how it’s calculated a few times a year. Currently, to the best of my knowledge, the popularity list is the accumulated sales of a book’s last 30 days compared to those in its category–but free books given away only count for roughly 10% of a paid sale, and price is factored in as well, in that the higher your price, the more each sale counts for on the list. Lastly, borrows aren’t counted as sales for purposes of popularity list rank. The formula looks something like this:
(sales + (0.1 x free downloads)) x (unknown sales factor) / last 30 days
A simpler way to think about it is gross revenue earned by your book over the last 30 days (with an additional boost depending on how many copies you’ve also given away). I’m not sure that’s a 100% accurate way to put it, but it fits the data we’ve seen well enough to work as shorthand.
In short, then, appearing on the bestseller lists is mostly all about having sold a bunch of copies in the last 24-48 hours. To appear high on the popularity lists, however, you need strong sales (or an extremely strong giveaway) over the last 30 days. Additionally, the higher your price, the fewer books you’ll have to sell to do well on the popularity lists; the lower your price, the more you’ll have to sell.
That’s a lot of information! What is the popularity list actually used for? It sounds like that’s what the changes are effecting, but, as a shopper, I wasn’t particularly aware of it until recently, so I never used it to find books. Do people actually browse through it? Or is it used for determining recommendations?
The popularity lists are pretty important. Obviously, Amazon has an almost endless assortment of ways to promote books from within the store itself, but I think the popularity lists are one of the major factors. See the main Kindle store page? With all those links on the left to a variety of different genres? Those bring you to the popularity lists.
So they’re pretty prominent. Both for browsing and, yes, for recommendations–when Amazon sends out emails along the lines of “You might enjoy these other books in Epic Fantasy,” the links they include take you to the popularity list for that category of books.
Of course, the importance of any given category varies quite a bit by its overall popularity with readers. Romance > Romantic Suspense might be just a little more important than Basketry > Underwater Basketweaving. Ranking high on the popularity lists of small categories won’t make much difference. But in the well-trafficked ones, it’s pretty big.
It’s hard to know just how huge unless you are actually Amazon, but if I were to make conservative guesses based on my experiences, being on the first page in Epic Fantasy might lead directly to 20-60 sales per day based on your visibility there alone. (And maybe much more. This will vary a lot depending on your book’s overall appeal. I’m sure A Game of Thrones benefits from it just a little bit more than my dinky indie title did.) In Science Fiction > Adventure, I’d say it might be good for as many as 30-100 sales. For the biggest categories like Romance and Mystery & Thrillers, the visibility the popularity lists provide to the top books might be responsible for thousands of monthly sales by themselves.
Key word “might.” This is really tough to estimate. But in my experience, a lot of people see these lists, both when they’re browsing around Amazon and when they’re directed there by emails. Based on post-free results from hundreds of different authors, I’m positive the popularity lists were the main drivers for the big sales Select authors used to see after making their books free. Now that it’s so much harder to achieve high visibility on these lists via free alone, I’m afraid Select authors are in for some much leaner sales.
Yes, in a recent Kindleboards post, you mentioned that Amazon’s changes would effect those using KDP Select. Can you summarize what’s been going on and what the changes may mean for authors?
Amazon has made significant changes to their popularity list algorithms twice this year. Around March 19, they started using three lists at once. Around May 3, they condensed that to a single list. The new list works as I’ve detailed above.
If you’re in Select and have been doing book giveaways, you may have noticed that you started selling fewer copies after a free run starting March 19. You’ve probably done even worse since May 3. That’s because free copies used to be weighted equally with paid sales on the popularity lists–which also looked at most recent sales most heavily.
But now that free downloads only count for about 10% of a paid sale, and the lists look at the last 30 days of sales rather than the last week or so, it can be really hard to land high on the popularity lists unless you give away a colossal amount of books. (Though if you can make it there, you’ll stick for longer.) Without the visibility of the popularity lists to drive your sales, you probably won’t see the “post-free bump” we grew used to in the first few months of Select. Select can still be an effective program, but for the moment, it’s far less useful for generating sales than it once was.
For instance, back in February, I gave away 9000 copies of my fantasy novel The White Tree. That was enough to put me at #1 on the Epic Fantasy popularity list for several days. I sold a lot of books! In March, I gave away another 4700 copies. On a similar version of the list we’re currently seeing, that was only enough to boost me to #65. I didn’t sell nearly as many books!
For a more in-depth look at these changes, check out my series of posts here, here, and here.
In that last post, you talked about how the new changes may make it harder for authors with 99-cent ebooks to rank as well. What exactly are you seeing and what price points seem to be favored?
Yes. Price now seems to be a factor as well. Collecting data on this is really hard–in fact, I can’t even say with total certainty this theory is correct–but there’s a strong correlation between price and relative position on the popularity lists. In short, the higher your price, the better you’ll place relative to your overall sales.
The favored price point in this new system is “as much as you can get away with charging.” It looks like $0.99 books have been pretty well massacred. $2.99 books can still place well (particularly when they’re boosted by giveaways), but they’re at a noticeable disadvantage. Something like $5.99 – $12.99 looks to be the ideal range at the moment. Affordable enough for people to buy in droves (if the quality is there), but with a high enough price to hang with all the high-priced traditionally published books.
This is not a call to jack up your prices. If you raise your book to $7.99 and only sell 20% of what you were doing at $2.99, you’ll be worse off on the popularity lists. And remember, the popularity lists are just one way to generate sales (although it is a significant one). But since price appears to be directly relevant now, it’s something to be aware of when positioning your book.
Any thoughts on why Amazon might be making these changes? To push people into their ideal $2.99 – $9.99 pricing bracket?
I don’t know. Could be, but it’s not like Amazon made any announcements about this. I don’t think Amazon builds these algorithms with overly specific goals in mind. Like, nobody in Seattle woke up one morning and said, “And now I ruin John Locke’s life! Ah ha ha ha!” As far as I can see, all they care about is what will make them the most money now and continue to do so ten years from now.
Do you have any parting thoughts on what these changes might mean for authors who hope to do well in the second half of 2012 and beyond? It seems like some of the “tricks” indies have used to outperform mainstream books (99-cent price tags, KDP Select free days, etc.) might not work as well in the future. Will this force us all (Big 6, small press, and self-published authors) to sell and promote our books in the same way?
This is just one more step in the ongoing and absurdly fast-paced evolution of the ebook market. The algorithms could change again tomorrow or six months from now. Amazon makes changes all that time.
That said, in the meantime, Select isn’t the money-printing machine it once was. To sell many books, you’ll have to do more with it than “set book free, sit on couch, drink fruity drink.” You need to have a secondary strategy to make your book visible after your free run’s over, or use your free run to specifically generate visibility for your other books. So maybe the strategy is to make the first book in a series free on a regular basis, or taking out an ad to run the day after your book reverts to paid to try to cluster as many sales into one day and climb the bestseller lists, etc. Indies are still in the process of working this out.
If these changes stick around long-term, we might see a convergence of prices, tactics, etc. between indie and trad publishers. But I think that, for better or worse, we’ll see yet another change before the year is up. Maybe several of them. While the current changes don’t look good for Amazon’s indie crew, we still have the advantage of being able to adapt faster to them–and to whatever comes next.
Thanks for taking the time to answer these questions, Ed!
~
Edward’s bio and information on his books:
Born in the Pacific Northwest, Ed recently moved with his fiancee to Los Angeles, where they’ve since accumulated three small furry mammals. A sci-fi and fantasy author, his short stories have appeared in a couple dozen magazines online and in print. His epic fantasy novel The White Tree and the postapocalyptic thriller Breakers are both available through Amazon. A more complete list of his work is available here.
Note: if you’re reading this on Friday the 18th, Breakers is free at Amazon, so make sure to check it out!