Google search data could predict which movies flop



The winter movie season is just getting up and running, so Google is taking the opportunity to remind everyone how much data it has on everything — including movies. In a newly released whitepaper, Google crunches the numbers on search volume and how it relates to how well a movie is received, and the results are impressive.

There are a lot of numbers to crunch out in Mountain View — movie-related searches are up 56% year-over-year, according to the report. Instead of going to a specific site, users interested in a movie simply Google the title and see what comes up. If you’re heading out to a film, you might Google it to see where it’s showing, or to look up reviews. Repeat millions of times, and that’s a juicy data set. Google has charted these title searches against the box office outcomes on a given weekend. Perhaps unsurprisingly, the data is a close match. Things get interesting when Google starts predicting the future.


In the seven days prior to a movie being released, Google’s research has found the number of search queries is highly predictive of how much money it will make. Forget all that Hollywood talk about buzz and timing — raw Google searches predict 70% of the variation in box office performance (the R2 value). That’s short of 100%, but this is a big data problem with a lot of moving parts, and this is just one metric.

Perhaps the most accurate predictor of a movie’s performance comes in the form of trailer searches. Google’s whitepaper points out that trailer searches on YouTube and Google four weeks prior to opening gives us a cumulative R2 value of 94%. That means 94% of the variation between how much two films make can be described by trailer searches. Google’s paid clicks are also useful sources of data, showing that a film’s ongoing performance can be predicted with 90-92% accuracy by the click rate.

The full whitepaper is available for download, but it won’t tell you what movie to see this weekend. You’ll have to Google around for that. I wonder what the search traffic on The Internship is looking like.