They Sure Gave Me Their 2 Cents Worth

In my last blog post, I mentioned that I wanted people to give me the worst-possible feedback about my books, because that will help me focus my attention where it is needed most. (It's not about being a masochist, really—I'd be happy if everyone thought my books were flawless!) All things considered, it's not surprising that my blog post asking for feedback did not actually generate any. That's OK, I didn't expect a rush of critiques.

However, I came up with a rather unusual idea. After pondering it for a good long time (30 seconds? maybe it was 45), I moved forward with it. I'm really glad I did. It turned out much better than I expected. Read on if you'd like to know what secret magical incantation... er, never mind. Read on to see how I used Amazon Mechanical Turk to get inexpensive and useful results.

In case you're not familiar with it, Amazon Mechanical Turk is a part of Amazon Web Services (AWS). It allows people to allocate large amounts of work in small slices to online workers. A typical example: Need a thousand photos tagged with useful keywords? Pay one cent per photo via Mechanical Turk and get the job done for ten bucks (plus a small percentage to Amazon). Without getting too geeky, this service can be run by software via the Mechanical Turk API, allowing for human-intelligence results that seem to come from a computer program. (This fits the history behind the Mechanical Turk name.)

For my needs, I didn't need to use the Mechanical Turk API. I didn't need to use anything more fancy than the AWS Web interface for building a work request. The joy I felt was not about getting results from a complex tool (that's a special geeky kind of joy), but in the results themselves. Going back to my last blog post, what do I want? Negative feedback. Who isn't giving it to me? Friends and family. Who would give it to me? Anonymous strangers. Cue the Mechanical Turk!

I spent a whopping 25 cents to get results that were insightful enough that I could take immediate action. One nickel went to Amazon for use of the service. The rest was divided among 10 people, and I basically asked them for their "two cents worth" about one of my books. For this first experiment, I chose Lesson One: Revolution! as an easy yet useful target. It has a higher price than others, but it doesn't sell well (a few copies a year if I'm lucky), and it has mixed reviews including some one-star ratings. I know it has problems, so I was prepared to get some pretty negative comments.

The task for workers was appropriately simple: Go to the page on Amazon for the book and take a look at the cover, price, blurb, and anything else that they might look at when book shopping. Then, assume that they would not want to read the book, and explain their main reason(s) for not being interested in it. I specifically did not say "buy" the book, just "read" it, because I didn't want money to be the focus of their response.

Workers using Amazon Mechanical Turk know that they have to provide a satisfactory product if they want to get paid, so I was protected against completely random nonsense ("asfasdfasdfwer" types of responses). In the work description, I also said I would reject any too-simple responses like "not interested" or "costs too much." Of the ten responses I got, only one was borderline (they basically said they don't read fiction), and I paid them their two cents anyway.

I'll post some quotes below, but the main value I received was realizing that I needed to change the blurb, which was something I could do without delay. I'd expected more comments about the cover, but only one worker said anything particularly negative about it ("seems rather ugly and boring").

Several workers talked about the blurb, and they helped me see something I did not consider before, that reviews and the blurb work with (or against) each other in a shopper's mind. I looked at them as separate entities, but—at least in that book's case—that's not appropriate. Maybe it's not appropriate for any book. It's easy for me, as someone crafting the various marketing bits individually, to think of them individually, but potential buyers will look at those pieces, and other pieces I don't create and can't control, together as one product.

After looking over the Amazon page for Lesson One: Revolution! here are what some of my workers said:

The description sounded interesting until it got to the 'romance' part, which made me lose interest again. The reviews indicate that it's highly political, but the reviewers can't seem to agree on what kind of politics it's representing. It sounds muddled and confusing.

...as soon as I saw the phrase "unexpected romance" [...] I was turned off...

The description isn't intriguing, it is dry and vague.

Another "gem" was seeing how the way I've presented the book can lead to people misunderstanding what it's about. For example:

I would choose to not read this book as it seems geared toward some of the fanatics trying to use isolated violent cases as a way to strip us of our gun rights. I believe in our right to own firearms and would not support a book or author who was trying to sway a person otherwise.

Those who know me personally will see right away that this is not accurate, I'm certainly not motivated to work against gun rights. One of the two-star reviews on Amazon for the book also suggests that the reader thought I was speaking out against gun ownership, so clearly I need to do a better job of disassociating the subject and themes of the book from gun ownership rights. (Although the book does promote personal liberty in general, it is not intended to advocate for or against the right of individuals to keep and bear arms.)

There are other valuable tidbits in the responses I received. Overall, for only spending a quarter and a few minutes to set up the "HIT" (Human Intelligence Task) on Mechanical Turk, I'm quite pleased with the results. I'm already considering repeating this for other books, as well as doing it again for this book to see how the new blurb compares against the old. Naturally, I'm still open to (unpaid!) constructive criticism from others, but lacking that I at least know where I can find people ready and willing to point out where my book presentation needs work.

After considering the input from my Mechanical Turk workers, I decided to scrap my blurb and start over from scratch. It has gone through several revisions over the years, but those minor adjustments were not accomplishing much. The new blurb now appears on Amazon, Barnes & Noble, Kobo, and elsewhere. Will it lead to more sales? Maybe not, but hopefully it won't prevent them either.


About the Author
Stuart J. Whitmore is an author of fiction and nonfiction, as well as a photographer, technology developer, and more. If you enjoy reading his blog posts, you might also enjoy reading his books. Take a look at the books by Stuart J. Whitmore today, and download your copy of one that looks interesting to you!