Results of the A/B split test I did on Facebook for my ebook cover astonished me. Read the factsPosted: March 3, 2014
It started as an attempt to find an audience for my new, non-traditional vampire novel, New Vampire Online. So far sales have been meager, but those who read this novel and posted reviews on Amazon really liked it—mostly 5 Star reviews.
So after asking best-selling author Cheryl Bradshaw for her opinion about running ads on Facebook and receiving a positive but tempered answer from her, I decided to move ahead with a week’s worth of ads which I could modify from day to day to see if I could pinpoint potential readership for the book.
What I ended up learning was much more than that.
The ad campaign was simple: Facebook asks the advertiser (in this case, me) to select a goal for the ad campaign. I chose the goal of click-throughs to the book’s page on Kindle. Then I moved on to a traditional A/B split test based on the two different versions of the book cover. Both had the same Dick Tracy-style illustration of Cate the flirty vampire. One cover, however, had a gray background, the other a bright red background.
(A digression here: When I initially published New Vampire Online a few weeks ago I used the gray background cover, then after a couple of weeks, switched, on Amazon only, to the red background. I liked the red version a lot. The artist who did the illustration liked the gray.)
These two versions of the artwork were the dominant elements in all four ads. Two ads with different color covers ran in the newsfeed, two ads in those little right column ads. The brief copy was the same for all, except for one link test I did later in the week. The targeting was also the same: Women, age 19-25, interested in the broad category of ‘Entertainment.’ The link in the ads went to the Kindle page.
On Day One, I foolishly let Facebook determine the cost per click and ended up paying $2.97 for one click to an ebook that’s on sale now at $1.99!! Nothing learned there, except don’t let FB determine the cost per click.
Cost per click now under control, Day Two’s results surprised me. The gray background cover beat the red background cover by better than 5 to 1 in the newsfeed ads. The click-throughs from the little right column ads were so minimal that the results were useless.
By Day Three it was clear that the gray cover was the huge winner. At the end of the day I began to make other changes, narrowing my search for potential readers to Women, 19-25, who were interested in Games, specifically online role-playing, action, etc. games. (A comment about role-playing online games from a young woman on Goodreads led me to this test. I thank her!) I stopped running the little square ads and paused the red cover ads.
By Day Four with only the gray cover ads running in the newsfeed to the new narrower target market, click-throughs increased. At day’s end I made an additional change: I did a version where the clicks went to iTunes. This was the only copy change made during the week. Click-throughs to iTunes surged ahead of the clicks on the link to Kindle.
On Day Five I tweaked the target market again, dropping the age to 16-25. Click-throughs sagged, so by the end of that day I went back to the 19-25 year olds.
I made no further changes on Day Six and the campaign ended on Day Seven.
So what did I learn from this testing?
1) The red version of the cover was clearly a huge obstacle that shunted potential readers/buyers away from the book. The gray version was on iTunes during the test and is now back up on the Kindle page.
2) New Vampire Online is apparently not a YA book. I thought of it as a humorous New Adult book when I wrote it, sort of a Sookie Stackhouse California style genre mash-up, and I now think additional testing for the YA market is not worth doing.
3) The book seems to have an appeal to female online gamers, but even after I reduced the amount I was willing to pay for each click-through, it still would not be cost-effective to use FB ads to generate sales. Click-throughs are one thing; actual sales are another, affected by issues other than the color of the cover.
4) I suspect that the heightened response to the ads linked to iTunes may be a function of Facebook viewership on iPhones.
So what started as a search for potential readers/buyers ended up giving me much more information.
My budget for all this? $30. And for those few dollars I discovered that Facebook is a terrific place for testing covers and learning more about potential book buyers. I can think of several other factors to test and if I do it, I’ll report on them.
Perhaps more publishers and indie/author publishers should consider doing this type of testing before they do ‘cover reveals’. It may be that the cover being revealed is actually damaging to sales, folks!
A final disclosure: I spent a number of years working in direct marketing and the lack of data in the world of book/ebook publishing astounds me. In any other field, there would be testing and more testing. Best practices and widely-known truths would be readily available based on hard data. I hope this test moves the ebook publishing industry in that direction.
Some shameless self-promotion: the day after I posted this information about testing on FB New Vampire Online received its 10th review on Kindle. 7 reviews are 5 Star. 2 are 4 Star. And one is a 2 Star–she admits she didn’t read the book. So now–ta-da–I can now advertise New Vampire Online on some of the major ebook newsletters. And I know for sure which cover I’ll use.