Market Research and What Publishers Don't Know

A New York Times piece suggests that book publishers largely fly blind, knowing nothing about their readers and so essentially publishing a lot of titles, hoping that some of them work out. The reason some get high advances is that the publishers think it will be a hit - but there is apparently little correlation. Numerical analysis does enter the equation:

In estimating value, editors rely heavily on an author’s previous sales or on sales of similar titles. Based on those figures and some analysis — about the popularity of the genre, the likely audience, the possible newsworthiness of the topic of the economy — they work up profit and loss projections.

Unfortunately, publishing success often hinges on something new that hasn't been a success in the past, so the publishers are effectively trying to drive a car by looking in the rear view window and seeing where they've just been.

Sounds like the recipe for an accident, and given the low profit levels of publishing, it has been. A friend pointed me to a blog entry by one of his friends, noting that the article made a big mistake by not talking to Amazon.com, because sure Jeff Bezos is sitting on a heaping pile of customer data. So someone - the online sellers - have a much better idea of what is happening in publishing. But even if they shared, the publishers would have to be willing to change the way they do business, and having dealt with a number of them as a writer, I'm not sure that they're willing. Suggest something new and the answer is almost inevitably, "But we've never done that before."