It has been hard to find good books that deal with the digital economy — especially in terms of microeconomics and strategy. The classic — Information Rules — is now almost two decades old. It holds up remarkably well but there is some embarrassment in still recommending it to students.
With their new book — Streaming Sharing Stealing — Mike Smith and Rahul Telang have finally produced the book I was looking for. This is a book you can give to MBA students that avoids hyperbole and teaches them how to think in the new economy. It is well written — I dare say a page turner — and it intermixes both anecdotes and proper empirical studies to provide lessons to managers.
The book is tag-lined “Big Data and the Future of Entertainment” and that tag-line is 100% accurate. This book does not deal with the entire digital economy but just with the entertainment sector. That is, music, TV, movies and books. And it has a theme: “you don’t use data properly, you know bugger all.”
Put simply, when it came to the digital economy, the reason these old industries got into trouble was not because they didn’t understand that change was going on, it is that they did not trust data analytics to help them make decisions. They relied on old assumptions, old suppositions that had never been tested. The reason they had not been tested was that they never had to. This is because entertainment is, and remains, a great industry. You win in a sector and you are one of a few oligopolists who can raise entry barriers and make money for decades. But the problem is, absent competitive pressure, you are not tested. And there comes a point when your decisions aren’t just a little bit off but are costing you half or all of your profits.
Relying on studies very familiar to readers of this blog (including many of their own), Smith and Telang relentlessly show how entertainment executives made mistake after mistake. Most of the studies are ones that look back and lay those mistakes bare. Others are based on closer relationships with the more enlightened companies that demonstrated before the fact that traditional assumptions were outmoded. And then Smith and Telang look to the future to offer new ways of organizing around data. That stuff I am less convinced of but you can see where they are coming from.
These studies are still going on. For instance, my own PhD student, Sandra Barbosu, has written an interesting paper (new version will be available soon) that examines how big data may have benefitted movie studios in the past. She takes Amazon Video (people who saw this movie also saw …) data and finds latent categories for movies that form clusters of similarity. So this isn’t just comedy etc but finer types of movie clusters. As it turns out, if you take those categories and apply it to box office data for the last two decades, movies that are part of latent clusters perform better than those who are outlines. (There are some endogeneity issues she controls for to make this more than some spurious correlation). The implication is that movie studios, had they had access to the data at least some companies have today could avoid movies that are outliers (not fitting into a latent category) and concentrate on ones that do. And we know that they didn’t have some intuitive sense of this because they produced movies that could be predicted (after the fact) to be ones that would have low box office revenue.
In summary, if you like this blog, you’ll like this book. I don’t have data to back that up but it is a solid theory.