My dissertation examined book sales and literally the only statistically significant predictor of popular books was marketing

Importantly, under-marketed characteristics included gender and race https://twitter.com/atrubek/status/1332836032249798656

There were eight characteristics correlated with sales, all marketing factors:

1) size of author advance (publishers couldn’t tell me marketing budgets but did say they were calculated with the size of the advance, which I tracked using PMktplc) r=0.4

2) fame of author r=0.4

3) association with something famous r=0.4

4) cover likability score r=0.4
5) number of Amazon reviews r=0.3

6) number of author’s Twitter followers r=0.3

7) starred reviews in specific locations r=0.3

8) YA/adult crossover ability r=0.3

When we look at marketing factors (minus Amazon ratings because they’re likely not causative) together, the r increases to 0.7

Here are some slides that have more detail, including likable covers

I specifically studied YA literature https://docs.google.com/presentation/d/1AYq8LVFP4mBxWGR5QxPZh3btoUCVQSvuVoQanZlSyZ4/edit

Once you have calculated a marketing score for every book, you can calculate how well a book should have sold and then measure the difference between how it actually sold.

I called this a resonance score.

A quantitatively notable disconnect occurred for books with high “race” scores (a combination of factors, including race of author, race of characters, and plot factors associated with race): their resonance scores were high

There was a similar disconnect for books demonstrably written to appeal to females

The YA adage that recommends you have a male protagonist so your book won’t seem “girly” is likely linked to an absence of marketing, which comes from implicit bias in both publishers & audiences