A declining cinema-going population and increasing competition from Netflix and video-on-demand are making it harder to monetize audience attention from the silver screen. One consequence is the wane of original content in big productions — studios preferring established franchises over truly novel, risky stories.

The criticism levelled at AI in this context goes: if algorithms get to decide which scripts are produced, wouldn't they exclusively allocate resources to derivative blockbusters from successful franchises? And if trained on previous films, how would they ever see the appeal of truly innovative content like Memento, Inside Out, or Being John Malkovich?

We believe this criticism is the result of confirmation bias. The tentpoles make the most noise marketing-wise — but other movies still get made, and the ever-growing importance of film festivals like Sundance proves that smaller movies are very much in demand. Furthermore, blockbusters are not necessarily that profitable in relative terms.

"The Fate of the Furious brought in over $1B worldwide but cost $250M to make. Conversely, Get Out was made on a $5M budget and has made approximately $255M — more than 50 times its production budget."

Use case 1: Absolute profit

Consider the top 10 most absolutely profitable movies from ScriptBook's database, released after 2000, predicted based solely on script. When focusing on absolute profit, we predict the "usual suspects" — sequels and films based on highly successful IP. This is expected: blockbusters win on sheer scale.

📊

Table 1: Top 10 most absolute profitable movies as predicted by ScriptBook

Upload the table image here

Table 1: Top 10 most absolutely profitable movies as predicted by ScriptBook based solely on script analysis.

Use case 2: Return on investment for indie films

Things become far more interesting when we turn to Return on Investment (ROI = net profit / production budget). We selected movies released between 2000 and 2015 with a production budget under $10M, and ranked them by ScriptBook's predicted ROI. The industry's actual average ROI across this set is 0.968 — meaning on average, these films barely break even.

📊

Figure 1: Actual ROI over top x% low-budget movies ranked by ScriptBook

Upload the ROI chart here

Figure 1: Actual ROI over top x% low-budget movies (production budget <$10M) as ranked by ScriptBook's predicted ROI.

The results are striking. If a studio had financed only the top 5% of scripts as predicted by ScriptBook, they would have won back 4.829 times the total production cost — far superior to the industry average of 0.968. Expanding to the top 20%, ScriptBook still beats the industry average by a factor of 2.

The top 5% includes critical hit A Separation and Fruitvale Station. The top 10% adds Safety Not Guaranteed and The Sessions. The top 20% includes Blue Valentine and the Oscar-winning arthouse darling Amour.

Conclusion

These experiments prove that ScriptBook's algorithms are perfectly capable of greenlighting smaller, independent, creative movies. Not only that — given a selection of movies, they manage to pick out the profitable ones far better than the industry average.

A lot of original movies are still being made, and they can be profitable. Get Out and A Quiet Place are recent proof that quality is ultimately rewarded. Making a profit doesn't mean raking in millions — but costs need to be recouped, and investing in a movie is a risky venture. AI doesn't pit Amour against the latest Fast and Furious. It might simply help make a difficult choice between two original low-budget movies — and make it better than a human would.

References

  1. The Numbers. Market data. the-numbers.com
  2. Bloomberg. Hollywood's 2017: moviegoing slumps to 25-year low. (2018)
  3. Box Office Mojo. The Fate of the Furious. boxofficemojo.com
  4. Box Office Mojo. A Quiet Place. (2018)
  5. Box Office Mojo. Get Out. (2017)