Using analytics to transform marketing processes in Hollywood

Social media becomes doubly powerful as it becomes both a platform for harvesting data and a vehicle for outbound communications and promotions.

Commentary - Hollywood knows how to market a movie, right? Sure, but there is always room to be better. In my previous article, ZDNet article on Connected Customers and Audience Analytics, I make the case that social media, big data, and analytics could seriously change the conversation about what insights are used to make marketing decisions.

Analytics in Hollywood marketing can use sophisticated tools and techniques to model the optimal places, channels, and frequencies to advertise and promote new content. Social media becomes doubly powerful as it becomes both a platform for harvesting data and a vehicle for outbound communications and promotions.

Here’s a few things that Hollywood marketers could do that would add incredible new insights to the decisions they currently make:

Audience targeting: Identify the target audience segments for specific movies, genres, talent, etc. to ensure that potential fans are reached by marketing efforts.

Creative & messaging: Adjust the creative and messaging of ads by learning about how fans are responding to movie previews, ads, and other marketing activities.

Influencers: Identify the core set of "influencers" for a movie/movie franchise to target them directly to get “word of mouth” out there.

Campaign effectiveness: Measure the social response to the ad campaign and change your course (see case below).

PR management: Get early warning on what's being said about the film or about a scandal, all of which can be used to increase awareness or temper bad press.

Marketing mix modeling: Social analytics can feed these marketing mix models to optimize marketing spend around marketing performance.

Competitive insights: Monitor competitors so you can make adjustments to your audience targeting, messaging, or marketing tactics.

Case in point:
This year a group of IBM consultants and research scientists worked with a major movie studio to perform social media analysis using a big data platform to discover what the consumer market was thinking about upcoming releases. The analysis harvested unstructured data from sources like Facebook and Twitter, electronically sniffing out over a billion relevant tweets, over five million relevant blog and forum posts, eventually narrowing down to ~3,500,000 relevant messages.

Our focus was on determining the real sentiment and buzz of two upcoming films as they were advertised over the Super Bowl season and then monitoring Twitter by the second during the game itself where a big advertising spend would be behind two trailers. We also tracked 100 other movies to understand the entire film market.

The level of buzz immediately during the TV trailers aired was significantly higher than normal levels – up to 20 times. Different movie trailers had very different buzz response levels from each other. For the Super Bowl, we saw significant variations between movie trailer responses, implying that certain films trailers were more effective than others in driving social response.

The level of intent to see a film varied anywhere from 10 to 40 percent of audiences that were buzzing about particular movies. However, during and immediately after the airing of the trailers, the level of intent increased or decreased significantly, showing the trailers’ effect on the percentage of people wanting to go see the film.

We’ve always felt that advertising changes the mindsets of filmgoers, but what a difference to know who, how, when, and how much the attitudes were changing by our marketing activity.

In our analysis we learned that there were varying levels of positive and negative sentiment about the films. Net “sentiment levels” ranged anywhere from -30 percent to +90 percent. During the period of the ad trailers being aired, these sentiment levels would change and would also increase or decrease for the period afterwards.

The category of sentiment also varied by film, meaning that a particular element of the movie resonated more than others such as cast, characters, plot, music, special effects, or the trailer itself. The level of sentiment toward a particular element of the film varied significantly, implying that audiences can react positively and negatively to different aspects of the same trailer. We could editorialize "representative tweets" and “evolving topics” in order to quickly evaluate what audiences were saying instead of reading through thousands of messages.

Finally, we were able to evaluate all these measurements across different social audience segments buzzing about the film, such as gender, geography, occupation, movie, and affinity groups, going as far as segmenting by region or even a persona type (e.g., comic book fan, book reader, etc.)

While all of this was fascinating data, the real impact was on the hard and fast marketing decisions that were made. For the movies we were analyzing, a particular line from the trailer sent the social buzz through the roof with excitement. We also identified the leading “influencer” segments, so we could target future media buys and tweak messages. For another, we learned that the trailer was missing a clear message and as a result, it would have to be redone.

The project was seen as a great success. As this line of innovation continues, I expect to see significantly more analytical insights being applied across the Hollywood marketing process.

biography
Richard Maraschi is IBM’s Global Business Analytics lead for the Media & Entertainment Industry where he helps some of the brightest media and entertainment brands solve their complex business challenges.