Can AI Predict a Movie’s Success?

Traditionally, the box office ruled. Big budget films would battle for key weekends scheduled around US national holidays, and movie executives would pour over the figures. A bad opening weekend would be very hard to come back from.

This year we’ve seen, perhaps sooner than we would have otherwise due to Covid, streaming services taking over from box office sales. The industry has been turned on its head, but has the usual film audience also changed? Gone are the days when you’d rock up to the cinema on a Friday night to pick from a handful of films. Now you’re at home, remote in hand, with a thousand of options at your fingertips. You don’t want to waste your time on something that isn’t quite your thing. So how does the film industry get savvy? They start by analysing their target market, and that’s where AI (artificial intelligence) comes in.

Although the technology used to make films has constantly developed, the way that decision makers analyse their market has been much slower. Having always relied on test screenings and their own experience of what makes a film successful, they’ve had to turn to smarter techniques to better understand what people want to see. AI may seem mind boggling to the uninitiated, but it basically means the study of computer algorithms that improve automatically through experience and the use of data. So what does that mean in the world of film, and how are production companies using it?

The movie business is in a state of flux right now, but this was happening before the world shut down in March 2020. Powerhouse Warner Bros. was struggling to create a hit in recent years and, in a move that infuriated some filmmakers last week, decided to release all of its 2021 movies on HBO Max on the same day they debut in cinemas. It shouldn’t have come as a surprise. Our viewing habits have changed and as a result the studio recently engaged in an analytics and financial modelling tool to help streamline their decisions about how, when and, crucially, where to release films.

The studios also rely on trailers as a key sales tool. Create a viral trailer and people will see the film. In 2018, 20th Century Fox teamed up with Google to analyse movie trailers using a tool called Merlin; their experimental movie attendance prediction and recommendation system. Using deep learning models that trained on customer data and film scripts to identify the basic patterns in audiences’ preferences, they were able to take out some of the guesswork. As well as deep learning and AI systems, companies can make use of machine vision systems to deliver immediate quality benefits and detect the undetectable.

Of course, streamers have been developing their software tools for years. Netflix have publicly claimed that their ‘recommender engine’ is worth $1 billion per year to the company. With thousands of movies and TV shows on their platform, the choice can be overwhelming (who else experiences the cursed Netflix scroll of doom?) Their tool takes countless data points to help make suggestions to users using machine learning. It’s the tech equivalent of browsing blockbuster on a Friday night.

Technology is changing the way we interact with the world and these tools will no doubt continue to be used to help the studios make better financial decisions. However, I want to see if the data will help move Hollywood into areas that desperately need their resources; getting more independent and middle budget films made and supporting diversity in the industry and in it’s stories. Only time will tell.