I sat down in a coffee shop in San Bruno, CA, with two even younger entrepreneurs than myself, Kurt and Dav. Google had just bought their company, Fflick, for a rumored $10 million.
Kurt and Dav wanted to help people figure out what their friends thought about new films that were coming out. The thinking was, film critic Roger Ebert might hate a certain movie, but your friends, who have similar taste to you, might love it. For movie studios, knowing whether Roger Ebert likes a movie isn’t necessarily a good predictor of box office sales. But knowing what masses of regular moviegoers were telling their friends about a particular film that was about to launch could help a studio adjust its promotional decision-making, perhaps releasing a different theatrical trailer or placing advertisements in certain areas.
Traditionally, a company like Fflick might have harvested the millions of monthly American moviegoers’ opinions through telephone surveys and paper ballots at theater exits. That’s an extremely complex and expensive operation. And that kind of research, by nature, tends to take a long time to collate and tally. By the time surveys confirmed that Gigli was indeed terrible, there wouldn’t be much a studio could do about it.
An ambitious company might try to overcome this challenge with money and muscle. It could throw staff at the problem, putting human surveyors with iPads in all the theatres. It could write a computer program to collate all of the data the surveyors input after bothering theatregoers, so people at the home office wouldn’t have to count the data by hand. It could implement innovations in its survey call center to squeeze more productivity out of its employees, so they could call more people per shift.
Or it could do what Fflick did, which, as you might at this point guess, was to use a lever to pull all that information out with a yawn.
Kurt and Dav realized that people around the world were already talking about what they thought of movies they saw. When you see a movie, if you’re like most people, you talk to your friends about it afterward, what you liked, disliked, whether they should see it, and so on. At the time Kurt and Dav were exploring their idea, the world had recently latched onto Twitter, the micro-messaging service we discussed earlier, as a way to share everything that was on its mind. Each day, millions of people updated their friends via Twitter on what online articles they liked, what music they were listening to, where they were going, and, in the midst of all the chatter, what they thought of that movie trailer for The Dark Knight. And not just what they thought of the trailer, but whether or not they planned to go see it.
So Fflick built its house on top of Twitter. They wrote code to suck data out, sort through it to find references to the names of movies (which Fflick sucked out of IMDB, a website that lists all past and upcoming film titles), and add up all the times a given movie name was mentioned with positive phrases like “can’t wait” versus negative comments like “don’t go” or “sucks.”
Kurt and Dav constructed a layer on top of the highest layer it could find. Automatically analyzing people’s comments for positive and negative reviews was difficult work, but it was less work than gathering the data themselves would have been.
“We spent about a month just heads down in the thing nonstop,” Dav said. “Then we were like, ‘All right, let’s turn it on, see what happens.’”
A week later, the Fflick team was driving to Los Angeles to meet with a dozen movie studios, all of which had seen the technology on launch day and begged the company to come talk business. They asked if Fflick could export its review data to their own programmers (and thus make Fflick a platform on which studios could develop their own software). Dav ended up programming the data platform (called an API) on the drive down. He coded it from the car, at one point holding his overheating laptop out the car window to cool it down.
Three months after that, Google absorbed the company and gave each of the Fflick team members a couple million bucks.