7. Mission
“Entertaining everyone & making the world smile.”
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
8. Company Shareholders care about?
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
Subscribers!
MORE SUBSCRIBERS!
Huge Competitive
Threat
9. Loyal & Growing
Customer Base
Reed Hastings cares about?
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
Content for everyone Accurate Recommendations &
Personalised Experience
11. Netflix
10%
Not
Netflix
90%
TV Viewership
Annual TV viewership: 1B Hours*
Number of TV households: 120 M*
Average TV viewing hours/day per HH: 8.33
Netflix users (2018): 66 M
Average Netflix viewing hours per
household per day: 1.65
In the US
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
Interesting Facts To Note
Source: Netflix ’18 Q4 Earnings Call
We can estimate
*Source: Neilson Research
12. Interesting Facts To Note
6000+ titles on Netflix (in the US alone)
In 2018 alone Netflix added 700+ titles
It could take upto 3.88 years to watch everything on Netflix
(if watched 24/7 without breaks)
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
13. THERE IS STILL MORE THAN 6 HOURS WORTH OF
OPPORTUNITY PER HOUSEHOLD!
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
14. Who are the users?
On the Dimension of Time Spent Watching Netflix
The Daily The Hardly Evers
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
The Weekenders
17. Survey
Sample Set includes:
100 Respondents
US based
Age group: 21 - 40
Equal gender representation
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
Aim: Validating if users experience fatigue on Netflix.
In person
interviews
Observing
users
18. Finding #1
User Behaviour:
Average Discovery Time (DT) for most users = 15 min+
0 0.15 0.3 0.45 0.6 0.75
Less than 15 min
10-15 min
30 min
More than 45+ min
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
Note: Total avg. viewing hours on Netflix per user: 1.65 hours
19. User Behaviour (when they don't find anything to watch):
Most Switch to another platform / Don’t watch anything
Finding #2
0 0.1 0.2 0.3 0.4
Quit to another platform
Quit and don't watch anything
Message a friend for recommendations
Watch something from NF recommendations
]
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
20. Finding #3
User Behaviour Frequency (when they don't find anything to watch):
Most users said they switch/quit frequently
Never
3%
Quite
Frequently
80%
Sometimes
12%
Almost
Always
5%
Note: Quite Frequently defined as: 2 / 5 or 40% of the times
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
21. Finding #4
0 0.15 0.3 0.45 0.6
Friends
Recommendations
Netflix
Recommendations
Own Curated
Lists
Social Media, Blogs,
Online Reviews
Users Trust:
Recommendations from friends > Recommendations from Netflix
]
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
22. 0 0.1 0.2 0.3 0.4
Quit to another platform
Quit and don't watch anything
Message a friend for recommendations
Watch something from NF recommendations
Finding #5
User behaviour (when they don't find anything to watch):
Most don’t text a friend in the moment
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
]
23. Summary of Company & User Research
More Titles +
Imperfect
Recommendations
Increase in
Discovery Time
User Fatigue
Drop
Off/Switch
• Netflix content is growing at a very fast pace.
• Users spend significant amount of total avg. viewing time on discovery.
• Netflix recommendations haven’t been perfect.
• Friends are better trusted source for recommendations.
Summary of the User’s Problem
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
24. My Conclusion
Since machine based recommendations are
great but not perfect, yet,
user fatigue is real.
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
26. Identifying the user fatigue point is crucial.
Adding trusted human aspect - at the points of fatigue
- to machine based recommendations can increase
viewership and decrease drop offs.
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
27. Meet Neil - The Explorer
Software Engineer, 32
Lives in Mountain View, CA
A Netflix Daily User
Loves watching & finding out about new shows &
movies
Also subscribes to Amazon Prime, Hulu, HBO
On weekends, he enjoys hanging out with friends
Pain Point:
“I watch anything and everything. But my Netflix feed has too much content
similar to what I watched but didn't like. My friends & I discuss interesting new
content all the time that I don’t see on my feed. A fresh set of unbiased
recommendations would be helpful at times when I’m just tired of scrolling.”
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
29. Don’t Snooze Yet – A Netflix Original
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
30. How will it work?
2 Key Aspects:
• Identifying unique fatigue point for every user.
• Leverages on ‘The Social Effect’ at fatigue point.
(Facebook is used as an example here)
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
31. Unique Fatigue Point
Estimated per user based on machine learning algorithms
built on data models that aggregates and captures signals*
as:
• Drop off time: Amount of time a user spends looking for shows before abandoning
the platform without playing anything.
• Viewing time per session: How long did the user watch the show, and did he
interrupt to go back to browsing.
• Discovery time per session : Amount of time user spends to browse shows before
watching something.
• Content Watched : What did the user end up watching, something we
recommended, something random and new, or shows he’s watched already but
repeats frequently.
*Based on an assumption that Netflix already captures the above usage logs
Understanding Netflix User Pain Points Proposing A Solution Measuring success Way forward
34. Measuring success
Discovery Time
Viewing Time
Drop-off Rate
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
At Product Level
35. Fatigue Detection
Don’t Snooze Yet Screen
Social Media Plugin
Measuring success
At Feature Level
Activation Retention
My List
How many times does Don’t
Snooze Yet screen show up
for Neil per viewing hour?
Did Neil drop off before Don’t
Snooze Yet screen showed up?
How many times does Neil play
from Don’t Snooze Yet screen?
How many times does Neil play
from what a followed friend
watched?
Has Neil integrated his
social media to Netflix?
Does he add friends’ shows
to his own list?
Is he adding more
shows/movies to his lists? Is
adding categories?
Is Neil interacting with Don’t
Snooze Yet Screen?
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
36. Launch
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
Beta Test MVP
Phase 1: Internal Test with Netflix employees
Phase 2:
Users who have agreed to Participate in Tests.
US based
100k users (standard Netflix testing sample size)
Don’t Snooze Yet Screen on TV only
Social Plugin enabled across all devices
37. Share your Netflix
stories with your friends now!
Expect lots of
memes!
• Social Media
• Push Notifications
• In-app / Web notifications
• Blog
• Email
Go To Market
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
Across all devices
38. Don’t Snooze Yet!
Share your Netflix stories now!
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
3 of your friends are watching Narcos Mexico
• Social Media
• Push Notifications
• In-app / Web notifications
• Blog
• Email
Go To Market
39. See what your friends are watching now.
Add their shows to your list. And Watch Better.
See what your friends are watching now.
Create your own recommendations. Watch
better.
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
• Social Media
• Push Notifications
• In-app / Web notifications
• Blog
• Email
Go To Market
40. Don’t Snooze Yet!
Share your Netflix stories now!
Adding recommendations from trusted friends to the Netflix
experience.
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
• Social Media
• Push Notifications
• In-app / Web notifications
• Blog
• Email
Go To Market
41. Don’t Snooze Yet!
Share your Netflix stories now!
3 of your Friends Have Been Watching Narcos Mexico, Neil!
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
• Social Media
• Push Notifications
• In-app / Web notifications
• Blog
• Email
Go To Market
42. Estimated Roadmap
Phase 1 (MVP): Don’t Snooze Yet
Product
Design/UX
Engineering
Sprint 1
Wireframe
Prototype:
1. DSY Screen
2. User Profiles, Social Plugin
3. My List Update
Sprint 2
Fatigue
Identification / Data
Model
Sprint 3
Back End/Server-side
Implementation
UI Development for all screens
Integration
Sprint 4
QA
Launch
User Testing; User Feedback
Coordination with
Marketing to prep for
announcement
Firefighting Across Teams
Test!
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
Design Validation
Beta Testing
44. Phase 2
Main Feed
Further down the road: Short Term
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
45. • Making fatigue point identification 100% accurate.
• Training data for better machine learning based
recommendations model.
• Capturing data on kind of content users prefer / don’t prefer to
watch.
Other Benefits
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
46. Further down the road: Long Term
Phase 1 Phase 2 Phase 3
Don't Snooze Yet
Main Feed
Pattern Based
Recommendations
Neil likes watching Netflix when he is getting dinner. He usually
picks something out from his Dinner Time shows. Recommend
something from there to him on his main feed.
Right recommendations to the
right user at the right time.
Understanding Netflix User Pain Points Proposing A Solution Measuring success & Launch Way forward
47. "I think about it really as us winning time away,
entertainment time, from other activities. Instead of doing
Xbox or Fortnite or YouTube or HBO or a long list, we want
to win and provide a better experience.”
- Reed Hastings
49. Hi, I am Sneha
Die-Hard Netflix Fan.
A Product Enthusiast
About me
sul.sneha@gmail.com
Editor's Notes
I’m going to talk to you about our beloved company - Netflix. Imagine this.
You come back home from a long day at work. Food is ready. Netflix is on.
You start browsing. You scroll through different titles. You’re on your list. 5 min. 10 min. 15 min. Food is over.
So let’s jump right in.
Netflix wants to: entertain everyone and make everyone smile.
Shareholders want: subscribers and more subscribers.
But we also care about: best content for everyone, beautiful CX with accurate recommendations leading to happy, growing customers.
For the context of this presentation we will look at
The time people watch shows on Netflix
The time between searching and playing a show on Netflix
And the number of times people leave the platform
Netflix takes up 10% of the 1B hours of TV viewership in the US - about 1.65 hours per day.
There is too much content on Netflix. Growing at a very fast pace.
A lot of opportunity to grow.
I decided to segment our users into these 3 buckets according to time they spend on Netflix.
With this in mind, I conducted a survey some in person interviews with current users. Focusing on the US market to start with. I found 5 things.
When asked what they do when they don’t find anything on Netflix to watch, most said they switch or quit.
Very few text friends.
Also, note. A lot of users are on 1 other platform at least.
With these findings in mind, I have come to the conclusion that…
A New Feature which we’ll call ‘Dont Snooze Yet’
So we are back to the browsing screen, 15 min in - fatigue is setting in.
And then this shows up.
A handpicked list of few shows that is fed in from your own several watch lists and from what your trusted circle of friends have been watching recently.
When you don’t know what to watch, this will be that friend you could but won’t text.
Let’s take a step back and see how this would work.
We’ll start with revamping an understated feature on Netflix - “My List”
You can now create several of your own lists - according to your mood, time you watch certain shows etc so that a specific kinds of shows and movies are easily accessible when you are looking for them.
When you add shows to your list, Netflix would aut-tag the shows, giving the lists their identity of genre so we can show you recommendations of lists of other people who watch similar genre.
You would be able to plug in to your social profile and follow your friends or even discover new people with similar tastes. You will be able to see what they are watching, there lists and people they follow.
Lastly, you could set your list or what you are currently watching to public or private mode also.
That is the first look of our brand new feature: Don’t Snooze Yet.
This aims to directly impact the discovery time and increase viewing time, decreasing the drop off rates. Better recommendations (with the ML + the human component) would lead to increased effective viewing time.
We will roll out the feature to our users and leverage on our strong social media presence.
Push Notifications.
Reach out via email too.
I am estimating about 8 weeks from start to launch for the MVP and will be working across Product, Design and Engineering.
If the MVP is a success, we will roll the feature out on the main feed itself as a part of phase 2.
The long term idea is to:
Use user usage and preference data to train our ML models better and make better content decisions and offer pattern based recommendations.
To conclude, this will be in line with our larger company vision and goal.