15% off membership for Easter! Learn more. Close

What parameters will you take into consideration while designing a recommendation engine for Netflix?

Asked at Google
1.5k views
Answers (2)
crownAccess expert answers by becoming a member

You'll get access to over 3,000 product manager interview questions and answers

badge Bronze PM

 

Netflix is an online on demand streaming platform. Its mission is to provide entertainment to its users.

The goal of designing a recommendation engine is :

1.       1. To increase the average viewing time per user by recommending relevant and entertaining content.

2. To decrease the average time spend per user in browsing which leads to frustration and disinterest.

Keeping the above 2 goals in mind, my approach would be to personalize the recommendations by creating user personas based on their demographics such as

-       location, age, gender, weather, day of the week, time of the day, content watched in recent past, content browsed in recent past, device used, content added to the watch list, shows based on genres liked/watched, actors liked/watched, languages preferred, average time spent per day watching Netflix,  average duration of shows watched, in case multiple users are sharing an account then what are the fellow users watching, watching habit – binge watcher, weekend watchers etc, keywords searched, previews watched, shows likes/reviewed, correlation between shows ex: user A watched shows 1 and 2 then user B who watched show 1 might also prefer watching show 2, Clicks per impression shown after the keyword search

Sample use case:

Persona 1: Young Millenial - A single male 25 to 27 yrs old living in NYC , earns 60k to 80k, loves watching latest releases (movies & shows) in action and drama genres on Netflix. Weekend Netflix watcher.

He turns on Netflix on a Friday evening, the preview shows the new thriller movie released today, he is also shown a list of shows which he has already been watching, there is a notification from a close friend asking him to check out the latest Jack Ryan movie that was added to Netflix this week.

Persona 2: Busy working woman – A woman 35 to 40 yrs old living in San Jose, CA, earns 120K to 150K, late night show watcher, interested in historical dramas, mystery and sci-fi shows. Shares account with partner interested in cooking shows and home grooming shows

She turns on Netflix on a Thursday night at around 9:30, a preview of the latest sci-fi show in shown to her, followed by a list of shows she has been watching along with a prompt for the arrival of new season of her favorite historical drama show which she had been awaiting for a long time. She is also shown a new recommendation for a baking show based on her watch history.

 

Access expert answers by becoming a member
0 likes   |  
1 Feedback
badge Platinum PM

Things you did well

  • Mission: Good description of the company 
  • Goals of feature (reco engine): You described a good list of goals of the feature
  • Parameters: Detailed list of parameters 
  • Personas: You listed user groups
Areas of Improvement
  • Structure: It was difficult to follow. 
    • Start with the companies mission, features mission and how it complements the overall Netflix mission. 
    • Asks clarifying questions and then revalidate your goal with IVR
    • List user groups and choose a particular user group 
    • List Parameters for this user group 
    • Mention metrics and choose at most 2 primary metrics to focus upon
    • Summarize
  • Clarifying questions: 
    • confirm with  the IVR if they want you define the potential solutions of the reco engine - how it is displayed to user/account (UI/UX), or criteria of how the content is recommended
    • Ask the IVR for clarifications if they have a particular user segment, geography, goal that you need to focus upon. 
  • Goal: Choose a broader goal such as engagement and validate it with the IVR 
    • you may also add that the engagement with recommended content assists netflix in understanding the user and create better content as well 
  • User groups: Try listing at least 3 user groups
  • Parameters: If the reco engine can predict the mood of the user based on past user data and overall mood in the geographical location, it would recommend the content that would have the highest propensity to engage
  • Metrics: mention some meaningful metrics that would speak to the success of the recommendation engine and the overarching goal of engagement
  • Provide short summary if there is time
Overall, I think your answer needs more content and structure. 
0
Get unlimited access for $12/month
Get access to 2,346 pm interview questions and answers to give yourself a strong edge against other candidates that are interviewing for the same position
Get access to over 238 hours of video material containing an interview prep course, recorded mock interviews by expert PMs, group practice sessions, and QAs with expert PMs
Boost your confidence in PM interviews by attending peer to peer mock interview practices, group practices, and QA sessions with expert PMs
badge Bronze PM

The purpose of the Netflix recommendation engine - is to provide personalized suggestion which reduces the time and frustration to find something great to watch

More clarity on personas/geography/ platforms used- Existing customers belonging to any geography and watching Netflix on any device 

Metric of Netflix while building a recommendation engine

1. Improve the average viewer watch time 

2. Increase the number of shows watched (considering viewers watched atleast 50% of the show)

Parameters to be considered while designing recommendation engine-

1. Watch history of existing user 

2. Genre / Category of the shows watched by user

3. Engagement with videos watched by the user 

  • Video liked/disliked
  • Video shared
  • Added to My List
  • Added in "Remind me" for upcoming shows/ movies

4. Language in which shows are preferred

5. Actors / Actresses whose shows are viewed by the user

6. Release year

7. Correlation of shows, such as user A watched show X and then watched show Y, similarly, user B has watched show X, may like show Y 

8. Shows watched in that geography 

9. How long the user watches any show (could be a possibility that user like small 20min tv shows)

10. At what time of the day user watches

11. Previews watched by users (its category/genre etc)

12. Keywords searched by the user

13. Clicks per impression shown after the keyword search

14. Shows watched by friends/family members sharing the Netflix account with multiple screens

Priority or ranking of recommendation based on Importance

As per my understanding, recommendation no. 1, 2, 3, 4, 7, 12, 13 & 14 would be more impactful in designing the recommendations engine for NetFlix.

 

 

Access expert answers by becoming a member
0 likes   |  
Get unlimited access for $12/month
Get access to 2,346 pm interview questions and answers to give yourself a strong edge against other candidates that are interviewing for the same position
Get access to over 238 hours of video material containing an interview prep course, recorded mock interviews by expert PMs, group practice sessions, and QAs with expert PMs
Boost your confidence in PM interviews by attending peer to peer mock interview practices, group practices, and QA sessions with expert PMs
Get unlimited access for $12/month
Get access to 2,346 pm interview questions and answers to give yourself a strong edge against other candidates that are interviewing for the same position
Get access to over 238 hours of video material containing an interview prep course, recorded mock interviews by expert PMs, group practice sessions, and QAs with expert PMs
Boost your confidence in PM interviews by attending peer to peer mock interview practices, group practices, and QA sessions with expert PMs