2. Types of data collected
Tracked every search made by the subscriber
Good or bad rating attributed by each subscriber
along with Nielsen rating.
ZIP code of subscribers
Type of device on which they streamed.
Number of times he paused the show; whether
exited the show before it ends
Subscribers switching off before/after credits began
to roll
3. Post Play Feature
Netflix Knew when the customer would most probably switch off.
Popup would then be carefully adjusted by the algorithm.
4. Personalized Recommendations
Recommendations given out based on:
• Prior viewing patterns
• Viewing behaviors and recommendations from
other users
• Based on the movie just watched
• Personalized Trailers
Results:
Customer ordered more movies
75% of viewer activity is based on these suggestions
5. Rating algorithm
• Netflix recommends movies based on ratings provided by the subscribers.
• Subscribers can rate even if they have watched the movie or not watched the movie.
0
1
Not Watched the movie
0.5 Partly Watched
Watched the movie
•Netflix also looks at data within movies. They take various “screen shots” to look at “in
the moment” characteristics.
A/B testing is carried out for testing
alternate webpage designs to find out
which design generates a positive result.
7. Integrating social media for the recommendations
After logging into Netflix using the twitter and
Facebook, subscriber updates can be analyzed by Netflix
to make context based recommendations to the users.
The research suggests there is different viewing
behavior depending on :
• the day of the week
• the time of day
• the device
• and sometimes even the location.
Designing the poster after analyzing the color recognition patterns of the users and finding the impact on
customer viewing habits, recommendations, ratings and the likes.
8. Challenges
• Analyzing relevant information
• Requirement of huge Storage capacities
• Providing multiple customization to different users with same account
• The authenticity of ratings provided.
• Difficulty in decoding the tweets/Facebook updates.
• Privacy of the customers
9. Benefits..
• Accurate Licensing fees based on collected data
• number of subscribers
• number of times they watched it
• Saving $40 million a show on marketing campaigns
Only 22% of
movies are
profitable
Only 33% of
new shows
survive for
more than
one season