The Secret Anatomy of KDP Select

KDP-Select_smallI’ve mentioned before the all-powerful shadow of Amazon algorithms, looming nigh incomprehensibly over any publishing endeavour. That this was not just a great hyperbole, I hope to show in this post, using the experience of four consecutive KDP Select promos in as many months.

1. What’s the Deal?

First, a bit of a background. If you’ve never tried to self-publish with Amazon’s KDP, you may not be aware of the Select program. It’s a program of exclusivity with Amazon’s platform; for 90 days you may not distribute your e-book in any other way. In exchange, your book can be borrowed for free by users of Amazon Prime (which grants a sometime hefty additional income, if you’re lucky) but more importantly, you get 5 days to give your book away for free.

Why would you want to give the work of your life for free? Well, because that’s the best possible way to game Amazon’s algorithms. “Game” is a harsh word here – “utilize” is better, since Amazon seems to have deliberately built their system around Select.

In fact, if you believe hearsay (and there is no other data than hearsay here. No company would ever divulge anything about their search algorithm) Amazon realized Select has too great a power over its market and worked to stem its influence a little. If you read self-publishing blogs from two, three, four quarters 😉 ago, a common theme is the reducing effectiveness of Select promos year after year. Perhaps we will see the last throes of the system this year – or perhaps not. Either way, contrary to what many authors report, the promo still works. It just doesn’t work miracles.

Select Chart 1
Sales by day since September 2012. Stars indicate promo days.
2. The Books Have to Move

I used to work in a brick&mortar bookstore many years ago. One of the first rules I’ve learned is that books have to move around. Even the greatest best-seller must one day be removed from the prime slot, to make way for a newcomer. There are many reasons for it, but the rule is sound, and all good stores use it. Somebody recently explained it in a perfect way – I can’t remember the source of the quote, please comment if you know:

“If you have a book like the Lord of the Rings, which will always sell 20 copies a day no matter what, and a new book which might sell 15 copies a day, there’s no point in keeping the Lord of the Rings on the best-seller display. You move it to the back, and let the readers discover the new book, instead.”

This is what Amazon’s algorithms do: they change the discoverability of the books around. They are the equivalent of the bookstore clerk. And just like the bookstore clerk needs to be told what books to put on which shelf, so do the algorithms need to learn about your book. This is where KDP Select comes in. A successful promo tells the algorithm: “this book has potential. Put it on display instead of that other one. Let the people see it.”

Select Chart 2
Cumulative Sales by Day. Stars indicate promo days.
3. What is a “Successful” KDP Promo?

Let me tell you first what a not successful one looks like. You will notice in the charts that my third promo – third star – had almost no impact on the sales. This was my benchmark: a giveaway lasting only one day, one that I almost didn’t mention anywhere on the social networks, and didn’t buy any ads for, just let it run its course. It was an abject failure, by any measure.

A successful giveaway must count in thousands. Three thousand is a good start. Ten thousand is better. Breaking into the best seller charts for free books is a must; breaking into Top 10 in your genre is great; breaking into Top 100 total is a guarantee of a long-lasting success.

As you can see from the charts, each giveaway resulted in bigger spike in sales than the previous one. But also, each time I gave away more books than the last time. But there is something else you can read from the charts, something that’s very interesting and tells you a lot about the power of the algorithm:

Regardless of how big the initial spike was, each bump eventually drops down to pre-spike levels. And fast.

Sales by Week since September 2012
4. We Know Major Tom’s a Junkie

The drops in sales are automatic, regular, unstoppable and easy to predict after a while. Once the algorithm asserts that your time in the spotlight is up, that’s it. The sales can halve overnight, without any apparent reason. And because your success was too quick to build any loyal following (see below) it eventually fizzles out without a trace.

This is a brilliant strategy – for Amazon. The Select quickly becomes addictive. The spike in sales is like a heroin rush, and the drop is like a withdrawal downer – with the promise of another rush as soon as you succumb to another exclusivity period. And of course, the strategy would not work if the program wasn’t so damn effective.

It’s all in the scale. The algorithm shows your book to millions of readers; there is no ad that reaches more people, no social network campaign. And yes, most of them will not be interested in it; others will just download it for the heck of it, and never read it. But a tiny percentage here equals a whole lot of people.  And as this last chart shows, this tiny percentage of readers will likely move on to your next book, and then the next; and the algorithms will pick up your other books and present them to other readers, and so on – the wheel will keep turning, slower and slower, until it grinds to a halt eventually, unless you go for another promo. But before it does, you will have sold more books than you could have imagined.

Sales of “The Year of the Dragon” saga, by volume. Volume 1 is the only one that’s ever been in KDP Select.
5. So What’s the Bad News?

So I suppose the only question that remains is: what are the negatives of using KDP Select?

There is one distinct disadvantage of this system. It misses the target. If your book’s demographics is broad enough, this may not be a problem; then again, if it’s broad enough, you may not need to use Select at all: your book is likely to sell on its own merit. 

But if you had a specific target in mind, then using Select is the equivalent of trying to shoot at ants with a double-barreled shotgun. Sure, you will hit a few people you wanted to read your book, but in the process you will reach hundreds who couldn’t care less about what you wrote.

If you want to build a loyal following, if you want to reach fans, Select is not the way. You have to do it slowly, in the old-fashioned style.

PS: There are at least two other ways to do what Select does, without exclusivity: use your friends and fans to “bum-rush” the charts, or pay for an expensive ad on one of the few remaining sites that reach thousands of eager readers. I haven’t tried any of these methods yet, but I’ve seen both of them work well.

All the above caveats remain, however: a flash-in-the-pan success is never a good way to build a stable following. It may only serve as a foundation for the real hard work.

You can trace the success of my last Select promo yourself, as between February 4th and 6th “The Shadow of Black Wings” is once again FREE on Amazon.

20,000 views! Big stats post

At some point today, this blog has passed 20,000 views! 🙂

The last mini summary was at 10,000 in November. It’s taken mere two months to double that – glad to see the numbers are still growing, instead of stagnating 🙂


The world map is looking roughly the same as before – there’s little left to improve. Africa and Middle East have filled up a bit, and there’s a lot of tiny island nations too small to show on the picture. The middle of South America remains strangely elusive – don’t they have the internet in Paraguay? Or is it all proxied through Brazil and Argentina?


The Top 10 countries remain roughly the same, with Canada climbing to the unexpected 4th. Taiwan is now 11th, ahead of India.


The bottom of the chart reads like a geography test. One can only guess what guided somebody from Swaziland or Federated States of Micronesia to my blog; and the single entry from China – was that a censor reading through the list of banned sites and clicking absent-mindedly?wolrdmap


Title Views
Home page / Archives 6,328
The Year of the Dragon 997
KDP Select – a week later. The price of one’s soul. 842
The Hollow Crown: Henry V – so strange I saw it twice. 808
New BBC2 trailer 724
How to be a successful… lottery winner 675
About Author 656
Parade’s End’s End. 474
Italo-Turkish War, 1911-1912 424
Formatting an eBook in 10 easy steps – part 1 422
Asylum of the Daleks – 5 questions, 1 observation 397
A different Vienna, part 2 – Gasometer: steampunk arcologies 367

If anything, the table shows how strong the convergence between traditional media and blogging still is. Four posts in Top 10 (discounting “system pages”) are about what I’ve seen on the telly. A recent breakthrough is the New BBC2 Trailer post – by far the most successful of the year so far. And it’s not even a proper post – just a YouTube link with copied lyrics. The KDP Select post, which I predicted to climb high last time, has done just that – now safe at No. 1

Spots 11-13 are occupied by more of my self-publishing ramblings.

I’m not posting top search results this time, as they are getting a bit meaningless and monotonous: roughly, equivalent to the Top 10 posts, just more jumbled.

Top referrers:

Referrer Views
Search Engines 4,543
Reddit 1,979 509 321
Facebook 292 263
Twitter 218
WordPress Dashboard 177 155
StumbleUpon 130

The big number comes from a one-day test of one of those paid referral places. Turned out to be just a bunch of spam-bots.

Search Engines, reddit, tumblr and FB are still big: Reddit is still massive, even though I think my site was ghost-banned for a while there 🙂 I’ve recently discovered that Google+ clicks often register as “search engine” query, so there may be quite a lot of these in the 4000 number. Passive Voice clicks come from just one comment on the blog. Twitter, Goodreads and StumbleUpon are trailing behind the big guys.

Top Clicks:

URL Clicks 374
Google 245 211 174 61 53 49 38 38
Google+ 3

The blog continues to serve as decent advertisement for my Amazon books: over 420 clicks combined. BBC clicks are almost all from that New Trailer post 🙂

Totals, followers, shares:

Content: 176 Posts, 176 Comments [sic!]

Taxonomies: 18 Categories, 716 Tags

Followers (includes Publicize) : 56 Blog17 Comments

Most views per day: January 15, 351 (discounting the experiment)

EDIT: Actually, my best day turned out to be today, at a staggering 606 views! 🙂

Most comments per day:  September 14 (the previous stats post – I have such a geeky audience 🙂

Most commented post: KDP Select – a week later. The price of one’s soul.

Comments per month: 13

So, what’s the next landmark number we’re aiming for? Dare I say, 50,000 ?

5000 views – viewing stats

Yesterday I broke through 5000 pageviews on this blog. That’s a cause to celebrate 🙂 The blog is active since the middle of June, so it’s taken almost exactly three months to reach the 5000. My other blog, that I moved from (and where all posts older than June come from) is at twice that much, but it’s been going on since July 2010 and Blogger counts pageviews a bit different than WordPress.

Here are some stats from these three months, if anyone’s interested. Warning: it’s a long post, and it’s got a lot of numbers. Feel free to skip it altogether 🙂 Continue reading “5000 views – viewing stats”