Despite the gloom and doom of some of the blog pundits, and despite the relatively weak effect of my last KDP Select promotion at the end of March, which came in the midst of Amazon’s shifting algorithms, I decided to put the two books in my Victorian San Francisco mystery series, Maids of Misfortune and Uneasy Spirits, up for another round of free promotions this month. While my goals have remained the same, my strategy changed in response to the changing algorithms, and, as a result, my outcomes this time around improved.
As usual, the primary goal for my promotions was to push both of my novels up on the historical mystery bestseller list and to get them as high as possible on the historical mystery popularity list. I have written numerous times about my conviction that keeping my books visible on these lists is a significant factor in my success. (Maids of Misfortune has been on these lists continuously since July of 2010, and Uneasy Spirits has been on them since it was published in October 2011.)
If my books fell off the top of these lists I would be dependent on driving potential readers to Amazon to look for the book. As a relatively unknown author with only a modest social media presence this is a difficult proposition. Instead, when my books are near the top of the historical mystery lists then people who are browsing these lists get the chance to judge my books by their covers, excerpts, product descriptions, “also bought” lists and reviews). Conversely, I have heard how dramatically sales decrease for books by other authors when these books fall off the bestseller lists.
As an aside, I don’t understand why some authors still argue that using free promotions devalues books. For example, the buyer can see a book’s fixed price (in my case my books are now $3.99––becaue like other indie authors I am feeling more confident about pushing my prices up from $2.99), so they know it is only temporarily free. I see the use of free promotions as the same as any promotion––for example, when traditional publishers pay (cut into their revenue) to get their print books onto the front tables of bookstores. I don’t recall anyone concluding “that they must not be very good books if their publishers don’t feel the books can sell themselves on their own!”
A secondary goal of doing another free promotion was to make the books visible on other categories besides historical mysteries, even if they didn’t stay there once the books went back to paid. The historical mystery category is a relatively small category (2,182 books), and I don’t usually sell enough books daily to show up in the top 100 of the larger categories like mystery–women sleuths (6,420 books), or historical romance (12,163 books), except during free promotions. In addition, I switched Uneasy Spirits from romantic suspense to historical fiction for its second category after the last promotion, and I hoped that this round of promotion would get it exposure for the first time in this fairly large category (22,000+ books). In short, this promotion would be another chance to expand my market beyond the historical mystery category.
By the middle of May, before the promotions began, Maids of Misfortune had slipped into the 7,000′s overall and 40′s on the historical mystery bestseller list. Uneasy Spirits was in the 9,000s overall and 70s on the bestseller list. Uneasy Spirits was dangerously close to dropping off the top 100, and was averaging 11 book sales a day, versus 25 a day in April and 42 a day in March (all these figures are for the US Kindle store). I understand that for many authors, 11 books a day would be nothing to sneeze at, but, again, if I want to sustain visibility I didn’t want to let that daily average slip any lower.
In case you haven’t been keeping in touch, Amazon apparently started testing new algorithms for its popularity lists in the middle of March (see this post by David Gaughran.) While these algorithms are secret it was very clear that a free download was no longer counting as a full sale. The effect of these changes was a drastic decrease in the post promotion sales bump most authors had been experiencing and fair amount of consternation among indie authors.
I confess I was relatively sanguine about these changes. Even though my own promotion at the end of March was seriously disappointing in terms of over-all sales, it did prop up my books’ rankings for a brief time and then slowed their decline. In addition, what I was witnessing was a very similar pattern to my post holiday sales from the year before––when KDP Select didn’t exist. Last year and this year my sales in April were 27-28% less than they had been at their peak in the 3 months after Christmas. The difference was that due to KDP Select the peak this year was ever so much higher than the previous year.
While Amazon’s introduction of KDP Select and its free days had given many of us a great gift in increased sales this past holiday, it was creating very volatile popularity and bestseller lists, and traditionally published books were being pushed further and further down those lists. It made sense to me that, given the DofJ settlement, Amazon would have a vested interest in proving that it could still provide a competitive market for those traditional books. Frankly it never had seemed right to me that some of my favorite historical mystery authors were doing so much worse than I was––it wasn’t their fault their publishers kept making so many bad decisions (high prices, bad formatting, refusal to participate in the Kindle Lending Library, ect.)
Then in the beginning of May the popularity list settled down––for now, and there has been general agreement that for the last few weeks the popularity lists reflect a new ratio where free downloads are only counted for about 10% (some say 5%) of sales for the purposes of ranking (ie 1000 free downloads =100 sales). There is evidence that the list is also weighted more heavily by a book’s sales (maybe even its total revenue) over the previous 30 days––rather than in the immediate promotion days. As a result, the effect of the already diminished download count is further flattened by the previous 30 days of sales averages. Edward Robertson has done a good job of summarizing the effects of these changes.
When I read the posts about the changes in the algorithm I decided to deviate from my previous strategy for free promotions. I had been putting Maids of Misfortune up with Uneasy Spirits at the same time for one day, then continuing Uneasy for a second day. My logic had been that Maids was my most persistent seller (and usually got its largest downloads the first day) and that people would see the two up together and a percentage would decide to go ahead and get both of them, boosting Uneasy’s downloads. The one time I had put up Uneasy by itself it hadn’t done well (and this was before the algorithm changes), and I assumed that people might be giving it a pass because it was a sequel. So each time I have promoted I kept Uneasy up for a second day, thinking that it needed the extra day to achieve a significant number of downloads and that people might have started Maids of Misfortune and enjoyed it enough to go back and get Uneasy the second day.
But, with the new information about the greater importance of the sales of a book during the 30 days before the promotion, I decided that I needed to rethink this strategy.
I wasn’t too worried about Maids of Misfortune. It hadn’t slipped down the rankings as far as Uneasy Spirits, and, because it was uploaded in 2009 when you could choose 5 categories it has a better chance of attracting free downloads. It also had 98 reviews, and I knew that this would help. I did decide, however, to leave Maids free for two days this time since it was going to take more downloads to achieve any sort of bump in sales with the new ratio.
More importantly, I also decided to put up Maids of Misfortune for free a week before I put up Uneasy Spirits (something I had never tried before). My thought was that if the free promotion of Maids increased the sales of Uneasy at all (and the ripple effect of free on sequels has been well-documented), then this would mean that at least 7 days of that 30 day average would have the increased sales to figure into Uneasy’s rankings––improving the chances that it would experience some sort of sales bump after it came off its free promotion.
So far it appears that this new strategy is working.
Maids of Misfortune was free May 19-20, a Friday and Saturday. At the end of the two days the book had 3206 free downloads in the US Kindle store. If the rumors about the new algorithm were right, this would translate into approximately 320 sales over those two days. The previous 30 days my average sales for this book had been 20 books a day, so not surprisingly these 2 days, at about 150 a day, did push up the book’s popularity rankings, which in turn increased the books sales and ranking on the bestseller list. The fourth day after the promotion Maids of Misfortune was in the mid 3000′s on the paid list, versus the 7000s where it had been before, and #12 on the historical mystery bestseller list, versus in the 40s.
And, during the free promotion for Maids, the sales of Uneasy Spirits doubled. The bump didn’t last past the promotion, but it does mean that Uneasy was in a slightly better position going into its own promotion, and that it has a slightly better 30 day average to help it sustain the bump it got from that promotion.
Uneasy Spirits was free May 25-25 (Saturday and Sunday) and did much better than Maids in terms of giveaways. While Maids only made it to 109 in the free store the first day and lost ground the second, Uneasy made it to #33 in the Free list and remained in the top 100 for the second day. (I suspect the fact that this was the beginning of the memorial day weekend might have caused Uneasy’ greater success). This meant it had much more exposure and achieved over 3 times the number of free downloads as Maids of Misfortune did (10,142 in the US Kindle store).
This of course meant an even bigger bump upwards for Uneasy Spirits when it went back on the paid lists since these downloads would translate into 1000 sales for the two days.
In fact, doing the promotions sequentially has benefited the sales and rankings of both books because Maids of Misfortune averaged 98 book sales over the 2 days that Uneasy Spirits was free, nicely adding to its 30 day average and pushing it up the popularity rankings as a result.
To date, 5 days after Uneasy Spirits joined Maids of Misfortune back on the paid lists, my average sales for both books is double what they were before the promotions began. Maids of Misfortune is now ranked 2945 over all, and it is #11 in the historical mystery bestseller list and #10 in the historical mystery popularity list. Uneasy Spirits is currently ranked 5138 over all and #22 in the historical mystery bestseller list (although 3 days after the promotion it did hit the 3000′s and was #18 on the bestseller list), However, perhaps more importantly for its long run sales, Uneasy Spirits is currently #8 on the historical mystery popularity list.
If Edward Robertson is correct in his analysis of the new algorithm, as the older, poorer sales for both books at the start of May drop off, and the newer higher sales during and since the promotions begin to dominate the 30 day average, both books should continue to do well in the historical mystery popularity list, which in turn should continue to boost sales and help maintain these books’ position on the bestseller list. In short, they may not have risen as high in the rankings as after previous promotions, but neither will they drop as quickly. If this turns out to be true, Amazon will have achieved its greater stability in the lists, but KDP Select will still permit indie books to be competitive as well.
My final point is that I learned about the new 30-day aspect of the Amazon algorithms on May 7th when I read about Edward Robertson’s blog discussion of the changes that had been made, and I was able to immediately respond (in the words of the title––change course quickly.) Four days after reading this blog post I had made my decision to give the free promotions another try, but this time sequentially, and I went onto my dashboard and scheduled both free promotions and began to make the arrangements to feature those free promotions on such sites as Kindle Nation Daily and Pixel of Ink. A week later the first of the promotions began.
I didn’t have to consult with anyone (like an agent or editor) or get the permission of a marketing committee to make this decision, the scheduling of the promotions took seconds, and the pre-promotional work I did took about an hour. As a result, I was able to move quickly to reverse the downward spiral of sales before my books dropped off the historical mystery lists and became invisible. I know that this is not how things would have happened in the world of traditional publishing where people keep using the analogy of how difficult it is to turn a big ship around to explain how slow the Big 6 have been to respond to the ebook revolution. And for that I once again give thanks for the power I have as an indie author to exert some control over the fate of my books, even when the winds of changing algorithms threaten to blow them off course.