December 16, 2015

Star Wars Box Office: The Problem with Predictions

Movie fans are abuzz over the box office records Star Wars: The Force Awakens might break. Many box office predictions rely on incomplete data methodologies that don’t help studios to maximize grosses, however-- and marketers should take note, as there are lessons to learn from the dark side of the marketing force.

If you haven’t spent the last few weeks living under a rock on Dagobah or trapped in the depths of the Sarlacc, chances are you know Star Wars: The Force Awakens opens this week.

You’ve probably also seen box office pundits debate whether the film will set a new record for first weekend grosses. Consensus estimates range from $180 million to $220 million, the latter of which would beat the $208.8 million standard Jurassic World set earlier this year. A few bold outliers have even predicted an opening frame of up to $300 million.

If the wide variety of estimates doesn’t inspire confidence, you won’t be surprised to learn that many of these predictions rely on pretty flimsy data methodologies. Worse still, the predictions miss the point. Box office prognostication is fun, but for movie studios, the goal isn’t just to create better predictions-- it’s to devise strategies that will help a film reach a wider audience.

This tendency to focus on the wrong variables and questions is relevant to more than just box office enthusiasts and film industry professionals. As we head into 2016, any marketer worth his or her salt knows consumer data is essential to intelligent forecasting and informed strategy. When film analysts mishandle data, all marketers can learn from the mistakes.

Here are four ways recent Star Wars box office analyses fall short and four lessons to help marketers resist the dark side as they become data Jedis.

Predicting the future isn’t important unless it also helps you change the future.

It’s enjoyable to predict how much The Force Awakens might make this weekend— but ultimately, the question doesn’t really matter. Instead of treating predictions as ends in themselves, here’s what we should be asking: What does the prediction teach us about what is or isn’t working, and how can we use this information to shape an even better outcome?

That is, knowing whether Star Wars is on pace for a given debut is less important than knowing why. Even for films that hit the zeitgeist, there’s always a bigger addressable market to go after. Last year, around 230 million people bought movie tickets in North America but since 2000, no single movie except Avatar - not even Jurassic World, The Dark Knight or The Avengers - has topped 80 million admissions. There are always a few million more people who might be persuaded to buy a ticket— if you can find them.

Rather than focusing on a given record or predicted number, then, studios must ask how they can identify and reach parts of the market that are tracking softly. This effort can mean the difference between a hit and a bomb, or between a solid success and a historic one.

The Takeaway: Even juggernauts have weaknesses. The biggest movie campaigns might appear to be firing on all cylinders— but there’s always weakness under the hood. Without doubt, at least one audience group is less passionate than the others. A successful prediction isn’t the one that’s right. It’s the one that identifies hard-to-find weaknesses and gives you the flexibility to course-correct before it’s too late.  

Social data is powerful but misleading.

Movie analysts have tried for years to derive connections between raw social media volume and box office results. This approach didn’t work back in 2006, when Snakes on a Plane infamously flopped despite fervent Internet chatter— and it doesn’t work today.

That hasn’t stopped people from trying. A recent Associated Press article, for example, notes that one month before its release, The Force Awakens had 405 million trailer views on Facebook and YouTube, 87% more than Jurassic World. Does that mean Star Wars is about to take a lightsaber to the dinosaurs’ numerous records?

Not necessarily. As YouTube stats demonstrate, trailer views are at best a diffuse indicator of box office potential:

  • 2014’s most-watched trailer was Fifty Shades of Grey, which opened last February and ranks only seventh among this year’s best debuts.
  • 2014’s ninth most-watched trailer was Dumb and Dumber To, which posted only the 27th best launch of the year and barely earned overall what Fifty Shades of Grey made in its opening weekend.
  • Interstellar’s trailer buzz ranked one spot lower than the Dumb and Dumber sequel’s but Christopher Nolan’s space epic still opened more than 30% higher.

Star Wars will likely open much bigger than any of the above-mentioned films— but needless to say, where a movie ranks in online buzz isn’t always where it ranks at the box office.

The Takeaway: Raw social volume is easy to count— but it’s also comprised of vanity metrics that aren’t actionable. Rather than treating social data as a direct predictor, learn which data points to ignore. Use the helpful data that remains to discover weaknesses among certain audience groups, and opportunities for better messaging.

Historical correlations are key.

A recent Hollywood trade report noted that The Force Awakens was the most-buzzed film on social media during 21 of 2015’s first 49 weeks. In contrast, no 2014 title held the top spot for more than seven weeks. This data indicates Star Wars will be a big hit— but no one doubts that. Comparisons that make obvious points aren’t particularly helpful.

To be fair, historical comparisons can help make sense of social data— but the comparisons have to make sense. For example, a studio might correlate a recent film’s box office with consumer language patterns on social media, revealing which consumer statements likely drove the financial performance. These correlations could help the studio to understand what social data says about similar films in the future. But if the studio omits the correlations or compares films that aren’t similar, the analysis falls apart.

The aforementioned report, for example, doesn’t include any financial correlations. It assumes one week’s social buzz is just as meaningful as any other week’s.

The report also relies on a particularly faulty comparison. The Fault In Our Stars was the 2014 film that led social buzz for seven weeks. It’s hard to see how the social performance of a teen-skewing, female-driven romance says anything about The Force Awakens, which arguably has as much mass appeal as any movie ever released. Indeed, The Fault In Our Stars earned only $124.8 million in its entire run— a fine result that was more than expected, but also a number Star Wars will obliterate by this time next week. Nevertheless, the comparison gets made.

The Takeaway: Few pieces of consumer data are intrinsically useful by themselves. Rather than analyzing data in isolation, use robust financial and historical correlations to find questions worth exploring and precedents worth comparing.  

Large samples aren’t always reflective samples.

Numerous reports have pointed out that The Force Awakens has demolished both IMAX and Fandango pre-sale records. Prima facie, these records seem like powerful predictors; after all, buying a ticket is a much stronger statement than streaming a trailer or posting a tweet. But again, this data just reinforces that Star Wars will post a huge opening—something everyone already knows.

Deeper insights are harder to glean. People who use Fandango don’t necessarily reflect the larger audience that will see Star Wars. As reports have pointed out, the majority of tickets are still bought in-person, with many people eschewing online sales so they can avoid online transaction fees.

If you’re an executive at Disney, it’s not enough to know that online sales are booming. After all, record-breaking Fandango sales were practically preordained the instant The Force Awakens was green-lit. It’s also important to find the audiences that aren’t buying so the studio can adjust its advertising to lure in holdouts.

The Takeaway: Don’t congratulate yourself just because certain precursors are strong. Does strength within one audience group mean a weakness among the others? Is precursor data over-representing a certain group?