Data Science and Skyfall: Mining the opening weekend

6th December 2012 4 Comments

What better way to demonstrate the real-time value of the connection between social data and marketing results that an analysis of the week leading up to the U.S. opening weekend of the latest James Bond film, Skyfall? Our data science team analyzes the social conversation to track marketing events surrounding the movie.

The recent launch of our re-syndication partnership with bitly allows us to see the connections between conversations and clicks like never before. Using simple analysis techniques to delve into rich data sets we are able to identify key influencer events, local variations in interests and which companies are best able to ride the wave of excitement around a breaking story.

When focusing on the week leading up to Skyfall’s U.S. opening on November 9, we discovered that people from 180 countries visited more than 2,300 different websites discussing Skyfall, covering reviews, trailers, interviews, the soundtrack and other angles.

The Big Picture

In the tracked window of time, click traffic to websites that hosted trailers and soundtracks remained relatively constant; whereas traffic to other sites saw sharp spikes. Read on for more about the traffic to movie critic Roger Ebert’s review as well as GQ’s Skyfall related coverage.


Often, large spikes in Twitter and Facebook traffic came at the same time but occasionally one of these referrers dominated as the following chart show.

As you can see from the chart below, 75% of clicks came from 8 countries while the remaining 25% were thinly spread across more than 170 countries. Unsurprisingly, most clicks came from the U.S. but given that Skyfall launched in Canada that same day and in Turkey seven days earlier, we were surprised that Turkey outranked Canada in number of clicks. Looking deeper, we discovered that the majority of Turkish links directed back to a Sony Facebook page but a Los Angeles Times article about the filming of stunts on location in Istanbul proved almost as popular.

The Global reach of local influencers

In our first chart we pointed out the large spike in traffic to Roger Ebert’s review on the Chicago Sun-Times website, so we wanted to look in to the cause further. On November 8th, Ebert, the nationally-renowned “two-thumbs up” film critic and journalist, posted a link to his review of Skyfall on both Facebook and Twitter. This quickly led to a burst of visits to the review by his followers, peaking at around 3 per second. These visits came from all corners of the U.S. (see below; darker orange = more clicks) but a few thousand visits originated outside the U.S., coming from Canada, the UK, and India—clearly demonstrating how traditionally local influencers can achieve a global social reach.

GQ Rides the Trend

By aligning itself with various aspects of the movie, GQ magazine was able to attract a lot of attention, comfortably ranking in the top five sites in our data set. Their interview with the film’s costume designer Jany Temime proved most popular. People also engaged with photographs of actress Berenice Marlohe and a review of the Bond watch.

Benchmarking Your Opening Weekend

Clearly, Skyfall has been a huge commercial success and a benchmark for buzz and brand-marketing. What we’ve looked at here merely scratches the surface of what’s possible; imagine having this data, not just for a single film, or opening weekend, but for the industry as a whole. You could:

  • Predict the commercial success of the film before it opens
  • Identify if you need to spend more to advertise or let the buzz build organically
  • Gauge the country-specific interest in the release and the influencers that are shaping the opinions

Given the millions of dollars movie studios spend on creating and promoting their releases, mining the social web for conversations and click streams provides a unique data set to help drive marketing spend and strategy.

  • Pingback: Data Science and Skyfall: Mining the opening weekend | DataSift Blog | Data Science and Big Data |

  • Pingback: Data Science and Skyfall: Mining the opening weekend | DataSift Blog | On Data Science |

  • mediaczar

    This 2010 paper by HP Labs might make good reading:

    In it, they go a step further and built a regression model for predicting the value of movies before their release (table 4 & fig 6).

    Since then, ironically, there’s been some research into whether the opening weekend is such a strong predictor of overall takings as it used to be. The argument (I believe) is that increased accessibility to / speed of publishing for peer reviews can sink a poor movie that’s been well promoted…

  • nimrodpriell

    Super interesting. However, the map of the US simply gives an approximate map of the population in these various states. I think a more interesting chart would give the # of clicks relative to the population of the state (or better yet, to some estimate of the measure population, e.g. twitter users) in the state, so that you can (perhaps) see interesting correlations.

Share This