Is There a Correlation Between Data Analysis and Hubris?

The fall of Ryan Kavanaugh and Relativity Media.

Two long-standing witticisms help set the stage for just how averse the film industry is to new ideas. The first goes like this:

“Not one person in the entire motion picture field knows for a certainty what’s going to work. Every time out it’s a guess, and if you’re lucky, an educated one.” – William Goldman

And the second:

What’s a great way to make a small fortune?

Start with a big fortune and invest it in the film industry.

Since the advent of the Hollywood studios in the early 20th century, reliably predicting and investing in box office successes has proven to be a fool’s errand. Any number of sure-fire strategies, equations, and harebrained ideas has been attempted, often with the backing of a new-to-the-scene entrepreneur with the charisma and marketing abilities to convince financial backers that he or she has finally cracked the enigmatic film industry. Invest now and profits are sure to follow.

Enter Ryan Kavanaugh, the young, brash founder of Relativity Media who often made that very claim – he had cracked the Hollywood code using incredibly detailed quantitative analysis that helped his company determine exactly which films to produce or help finance in order to achieve sustainable profitability. Kavanaugh enthusiastically stated that Relativity’s goal was to “take advantage of and help fix inefficiencies in the market. Media businesses in general are built on protocols that, in some cases, were established 100 years ago. The entire film business is still run around theatrical, yet theatrical is only 20 to 25 percent of the pie. While there’s a big correlation, box office certainly isn’t the judge of profitability. I’ve looked at the film business today, and I’d like to think that we’ve helped make it a more profitable business—for us and a lot of other people. ”

Relativity Media’s fundamental claim is that it understands the reams of data pouring out of the film industry better than anyone else in town. When deciding whether to back a film, the company uses a series of regression analyses pulled from the box-office and other performance metrics generated by every single film released during the previous 10 years. The company slices the data for as many variables as it possibly can in order to divine which variables actually imply correlation with a film’s overall profitability. It then uses this information in determining which films the company should produce, with the aim of generating industry leading returns on invested capital due to film out performance.

And, at first at least, it appeared that Kavanaugh may have been on to something. Early investments in Talladega Nights, Immortals and others seemed to support Relativity’s unique ability to pick lower-budget films capable of minting box office gold. But ultimately those successes proved to be the exception and not the rule.  According to Forbes magazine none of Relativity’s films had earned even $70 million at the worldwide box office, a pittance in today’s box office and likely well below what it cost to make and market most of the films, leaving Relativity with substantial losses. Given this string of flops, the company was forced to declare Chapter 11 bankruptcy this past July after the company failed to repay ~$320 million in debt that had come due and tens of millions in bills from vendors. Kavanaugh has recently announced plans to purchase Relativity out of bankruptcy and resurrect the company’s business model but he has a long way to go before proving that data analytics can be used to predict box office successes. While many continue to believe that movie business remains overly reliant on antiquated norms and “gut instinct”, at least in this instance it appears that data analysis was more useful in luring in investors than it was at picking box office winners.

 

http://deadline.com/2015/10/relativity-media-bankruptcy-case-explain-1201570811/

http://variety.com/2015/biz/finance/relativity-media-reorganization-plan-1201644175/

http://www.bloomberg.com/news/articles/2015-07-31/would-be-hollywood-mogul-backed-by-burkle-and-mnuchin-bombed-out

 

Previous:

Pymetrics: Play Games, Get a Job

Next:

Netflix: Bringing Data Analytics to a Creative Industry

Student comments on Is There a Correlation Between Data Analysis and Hubris?

  1. Great Post!! In my post I make the counterargument that analytics can lead to a long term competitive advantage and use Netflix as an example. Would be great to know your thoughts on the same.

  2. Having never heard of relativity media before your post, I found the company’s proposed value proposition to be quite interesting. After all, one would think that with the right analytical tools, 10 years of box office data could prove useful in creating a formula for a tent-pole film. However, perhaps the flaw here lies in the fact that the film industry is a two-sided marketplace and any proper evaluation of a films success should be measured in accordance with changing consumer tastes, trends, etc. If Ryan missed this piece in his analysis, then perhaps diligence is more appropriate to blame than hubris.

  3. Yeah, great post and thoughts about this company. It seems like causation is really difficult imply from quantitative variables that go into movie production – leading to Relativity’s big budget flops. It’s often the intangible details that make a movie go viral – some character that is especially likeable or chemistry that works well. Also, two very similar movie concepts could have completely different reactions among moviegoers based on small differences in marketing, timing, actor mix, etc. Thanks for the info!

  4. Thanks for sharing your thoughts on this topic – great to read this side of the argument as a complement to some of the posts that focus on Netflix and Amazon and their approach to content creation. I think pointing out the importance of the human element and judgment in artistic creation, as you allude to, is difficult to replace entirely with data and algorithms – both in artistic creation, and in reviewing art. I think this type of lesson can serve as a tale of caution and extend to the danger of over-reliance on data and algorithms in other fields too – no matter how much data we might have, our ability to properly interpret and use it will always need to be paired with the right amount of human judgment.

  5. Thank you for the article. Based on your analysis, do you suggest that a mathematical algorithm cannot be derived from historical data to predict smart investment in films? It appears that irrational and human-specific input lead to certain investments as well as box-office success. I am also curious how digitization disrupts the film industry by inticing consumers to not go to the movies. How do companies like Netflix further steal market share from the industry?

Leave a comment