Amazon’s Ever-Changing Algorithms Part 2: Oops, They Did It Again

So not 12 hours after my post about how Amazon’s algorithms work, Amazon changed their algorithms again. This is a big enough deal they apparently just caused me to quote Britney Spears.

Fortunately, the latest changes weren’t a complete revolution. By all accounts, there is once again a single list seen by all customers. I’m not sure exactly how this new list works–the Avengers are still working on it–but it seems to hew very closely to one of the lists we already understood. And if you are a Select author who leans on free giveaways for sales, here is my current advice to you:

Sorry, couldn’t resist. I actually don’t think we should flee the battlements just yet, for reasons I’ll get into lower down, but this is bad news for Select authors. Phoenix Sullivan has alluded to this, and will surely have more to say herself, so keep watching. Anyway, cause for panic: the new list looks an awful lot like List B. As a refresher, here are the main mechanics (that I am aware of) for how List B works:

• Ranks are determined by the last 30 days of sales, with no extra weight given to the most recent sales
• Free book downloads are discounted heavily–maybe as little as 10% the value of paid sales
• Borrows don’t count as sales

I’m not sure about whether borrows count or not on this new list. Won’t have the data on that front for a while. I am positive free books are counted, and that they’re counted at a discounted rate–feels like 10-15% of a sale. (In other words, for every 10 freebies given away, you’re credited with 1 sale for the purposes of pop list rank.) You can see this for yourself by trolling the popularity lists. You’ll see a few titles that are permanently free scattered across the first few pages. If freebies didn’t count, you wouldn’t be seeing them on the lists at all, and if they did count, you’d see a lot more permafree titles and Select titles higher up on the lists.

Also, if a book has fewer than 30 days of history behind it, as in it’s a new release, that doesn’t seem to be counted against it. Seems like the pop lists will just count however many days it does have on its record.

What does this mean, then? Well, for starters, it’s probably the end of the 3-day bump. This was the term coined on Kindleboards for sales on Select titles that had recently been free. In the past, List A would update roughly 40 hours after your book reverted to free. It counted freebies to sales at a 1:1 ratio (or very close to it) and weighted a book’s last few days of sales very heavily. So if you went free and gave away 5000 books, in the afternoon two days after your promo ended, your book would be credited with 5000 sales, vaulting it to the top of the “most popular” lists. With your book in front of so many customers, you’d see a lot of sales, spiking late on Day 2 and carrying through Day 5 or so as your rank decayed and your book was pushed down the lists by new titles rolling off free.

I don’t think that’s going to happen anymore. For one thing, List B isn’t weighted toward your last few days. It seems to take an average of your last 30. That levels the playing field for books that have been selling steadily for the last month while largely ignoring sales spikes that don’t prove to be lasting. For another thing, when you come off a giveaway of 5000 books, you’re no longer credited with 5000 sales towards determining your pop list rank. Instead, you’re credited with something more like 500. Possibly 750, or even 1000–I’m not sure just what the conversion rate is–just that it’s definitely a lot worse than 1:1.

If you don’t get credited with sales, you won’t vault up the pop lists. If you don’t vault up the pop lists, you won’t be seen by customers. If you don’t get seen by customers, you won’t sell books.

Does this mean those of us in Select should flee from the incoming flood of orcs? Perhaps. Everyone loves a good stampede. But even if the three-day bump has departed for the swift shores of Valinor, freebies aren’t worthless. Giving away a whole bunch of books still means you’ve got your book in the hands of a whole bunch of readers. I hear word of mouth is a thing. While your book’s free, you’ve got tons of attention, meaning some of that attention will bleed over to your other books–especially if it’s a series. If you don’t have a built-in fanbase eagerly awaiting your next release, it might be a viable strategy to put it in Select, make it free right away, and get it some alsobots to help prop up early sales and build a little bit of what the kids call buzz. Furthermore, if you give away just an amazing amount of books, it may still be enough to pull you some decent visibility. Even if you can’t make page 1 of the pop lists anymore, there would be value in using freebies to push yourself up to page 5.

Also, this is just how things look right now. As should be obvious, Amazon changes how they do things all the time. They are wise. They want to make money. If they still want Select to succeed, at some point they will do something to help it, either by changing the algorithms again or by adding new perks to the program. What we’re seeing now will probably be completely irrelevant in another month or three or six.

It should further be noted that I don’t have the algorithms worked out to a T. I know what I’ve described here is incomplete. With the help of the Avengers, I hope we’ll know more soon, but let me stress this: I don’t know everything. I could be wrong about some parts of this or about all of this. I could be a brain in a jar dreaming that Amazon just screwed me!

But I’m pretty confident about what I’ve laid out here. Confident enough to make it public. For those of us who’ve been leaning on Select to sell books, the next few weeks or months could be lean ones. Plan accordingly.