Big Data Day LA 2016/ Data Science Track - Data Science + Hollywood, Todd Holloway - Director of Content Science & Algorithms & Conor Dowling - Content Analytics Manager, Netflix
Netflix will spend six billion dollars this year on content, making the company a major player in Hollywood. An increasing portion of this spend will be on original shows such as House of Cards, and original movies such as Beasts of No Nation. As we continue to expand our involvement with Hollywood, we want to leverage data and data science to make the best decisions possible. This talk will explore areas where we see the most opportunity to apply data science to Hollywood, and some early approaches we've taken.
Similar to Big Data Day LA 2016/ Data Science Track - Data Science + Hollywood, Todd Holloway - Director of Content Science & Algorithms & Conor Dowling - Content Analytics Manager, Netflix
Similar to Big Data Day LA 2016/ Data Science Track - Data Science + Hollywood, Todd Holloway - Director of Content Science & Algorithms & Conor Dowling - Content Analytics Manager, Netflix (20)
"Impact of front-end architecture on development cost", Viktor Turskyi
Big Data Day LA 2016/ Data Science Track - Data Science + Hollywood, Todd Holloway - Director of Content Science & Algorithms & Conor Dowling - Content Analytics Manager, Netflix
4. Timeline
1999 DVD subscription service starts
2007 Streaming service starts
2012 Netflix adds a studio
2016 Netflix goes global
5. DVD Service
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
6. DVD Service
Content Buying: Nonexclusive, and costs are not distributed across users.
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
10. Streaming Service
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
Content Buying: Often exclusive, and costs are upfront and shared across users.
11. Studio
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
Content buying creation: Exclusive, global rights. Creative freedom.
12. Studio - Series
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
2012 2013 2014 2015 2016
+ 3 more + 2 more + 20 more + 30 or so more
13. Studio - Docs
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
2012 2013
+5 more
2014 2015 2016
+5 more +many more
14. Studio - Films
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
2015
2016
+ 8 or so more ...
15. Studio - Talk Show
1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
2016
16. 1999 DVD service
2007 Streaming service
2012 Netflix studio
2016 Netflix goes global
26. Can we use data science to create content?
Answer: Netflix is not doing this. In fact, it’s the opposite,
we want to give create freedom to the creatives.
27. Can we use data science to help select content?
You bet. Ultimately all decisions are made by experienced buyers, but models can provide color.
29. What makes content valuable to Netflix?
● How many hours viewed?
● How many people finished watching?
● New subscribers added?
● Awards?
● Acclaim?
● Binge worthiness?
● Cult appreciation?
(no advertising)
30. We choose ‘efficient content’.
value / cost > 1.0
What makes content valuable to Netflix?
31. Licensing - Hollywood
600 films and 400 scripted series are made per year in Hollywood. The number of films hasn’t changed
too much over time.
Major Film Studios
32. Licensing - Hollywood
400 scripted series are made per year in Hollywood. Scripted series has grown from 200 to 400 since
2009. The quality has gone up during that time as well.
33. Licensing - International Studios
https://stephenfollows.com/how-many-films-are-made-around-the-world/
34. Licensing - Using Machine Learning
Available
Titles
Demand
Features
Demand
Predictive
Model
Adjust
If efficient,
make an offer
35. Licensing - Using Machine Learning
Available
Titles
Demand
Features
Demand
Predictive
Model
Adjust
If efficient,
make an offer
E.g.
● Past performance on Netflix (if previously licensed)
● Past performance of similar titles on Netflix
● Broadcast ratings
● Theatre ticket sales
● IMDB scores
● Talent involved
● Reviews
● Awards
36. Licensing - Using Machine Learning
Available
Titles
Demand
Features
Demand
Predictive
Model
Adjust
If efficient,
make an offer
37. Licensing - Using Machine Learning
Available
Titles
Demand
Features
Demand
Predictive
Model
Adjust
If efficient,
make an offer
E.g.
● Buyer judgements
● Deal term adjustments
42. Originals - Where do ideas come from?
2001 Book optioned before publication
Samuel Jackson attached to lead role
2007 Appears on blacklist
Christian Bale attached to lead role
2011 Funded $60M
Leonardo DiCaprio attached to read role
2014 Filmed at cost of $135M
2016 Wins best director & actor
43. Originals - Using Machine Learning
More difficult problem than licensing
● Less data
● Moving target - ideas and scripts can evolve
● Fungible - execution varies with talent and budget
44. ‘House of Cards’ script = X
Those
members
also watch:
Originals - Script Similarities
51. Next Inning...
How do we choose the best mix of content globally?
Can we identify valuable content earlier?
Can we assist the creative process (e.g. talent selection, location selection)?
52. Thank you
We’re hiring in Hollywood!
Sr. Data Scientist (https://jobs.netflix.com/jobs/860491)
Sr. Data Analyst (https://jobs.netflix.com/jobs/860765)
Sr. Design Technologist (https://jobs.netflix.com/jobs/860770)
Sr. Data Visualization Engineer (https://jobs.netflix.com/jobs/860564)
Sr. Data Engineer (https://jobs.netflix.com/jobs/860711)