Netflix's Data Dilemma — Hao Nan Peter Zhang

UX Research

How much privacy are Netflix users willing to give up for personalized services?

Project Overview

Background

Netflix, the largest online streaming service in the world. Its business model is a subscription service that offers personalization recommendations to help users find the shows interest to them. By leveraging a mix of user-provided interests and ML algorithms, the service provides recommendations in various formats to users. With the growth of digital platforms, privacy concerns have caught the attention of many people, especially in streaming services.

On the one hand is the joyful experience Netflix provides, on the other hand companies are collecting millions of user data everyday, raising very real ethical concerns related to personal privacy. In this research, we will be focusing on the trade-off that Netflix users encounter: the decision between disclosing personal information in exchange for personalized services.

Project Details

Peter Zhang, Eric Cheng

Applied UX Research Final Project

4-Weeks

User Interviews, Miro, PowerPoint

Team:

Type:

Duration:

Tools:

Research Target

Netflix users in their 20-30s

Research Impact

Improve Netflix service by resolving potential concerns users have, increase the retention rate of existing customers who have concerns around its service related to privacy

Increase Netflix’s reputation among business competitors by being one of few to take actions on user privacy issues, increase the reputation among potential users for customer growth.

Research Question

How do Netflix users in their 20s to 30s perceive sharing their data in return for personalized services and what impact does this have for their continued use of the service and why?

Hypothesis

Adhering to the privacy paradox theory, we believe that while young professionals may have worries regarding the transparency of their data utilization, they are still inclined to trade their information with Netflix in exchange for more personalized service.

Qualitative Research

User Interviews

Conducting interviews can assist us in learning users’ attitudes towards exchanging personal data for personalized services. This qualitative method is excellent way to gain insights into our target user group and develop empathy toward their perspectives.

11 participants from 9 countries of origin

Reached out through personal connections

Hour-long semi-structured interviews

Our biggest limitations are our sample size and sample diversity. Since we are focusing on Netflix users in their 20s to 30s, we would ideally like to have more than 12 participants from that age range.

We only reached out to friends and acquaintances due to the time and budget constraints. As a result, all our participants have received at least a bachelor degree and are educated mainly in the design/tech industries . From our 11 participants, there was also an unintended dichotomy in the users’ age, cultural background and cultural environment which we acknowledged our research findings. We recognize this dichotomy divides our user profiles into two distinct categories and requires further research in order to validate how these insights would scale.

Sampling:

Recruitment:

Format:

Limitations:

Affinity Mapping

We used thematic analysis to organize quotes and responses into categories. We then mapped out the responses by questions and hierarchy into three major sections: how users select what content to watch, their attitudes towards data collection and their utilization of Netflix’s recommendation feature. Through the affinity mapping exercise we were able to really emphasize with our users’ preferences and motivations and allow us to start crafting our user personas.

“I have to accept the fact that no matter what I use, if I am online, my information is being collected.” -Participant 5

Quantitative Research

User Survey

With insights gathered from our user interview, we launched a secondary survey to find out if user attitudes towards privacy and recommendation applied at a larger scale as well. We gathered over 90 responses and this helped us validate findings from our interviews.

We asked the following questions:

Do you consider giving away your personal information in exchange for Netflix’s recommended content a costly decision?

What is your interest level in having more control over the types of shows gets recommended on Netflix?

Are you interested in exploring your watching behaviors on Netflix?

Analysis

Combining data from our interview and surveys, we were able to concluded that:

Netflix users in their 20 to 30s consider sharing their data to be acceptable and a necessary aspect of the platform’s personalization service. Their primary concern is the amount of value they receive in exchange for their data, and they are presently dissatisfied. This disappointment with the service could influence their decision to continue with the service.

To elaborate on the disappointment users face we crafted two personas:

Stacy

Age: 24

Occupation: Graduate Student

Stacy is finishing the final semester of her graduate program . She has been living by herself for the last two years and is slowly developing her independency. Although she has a good sense of who she is as a person, she’s still eager and curious to explore her tastes and interests.

When it comes to Netflix, Stacy has been subscribing for 5 years now and watches on average 8-9 hours a week. She chose Netflix mainly because it has most of the popular shows and shows all her friends are watching. She likes to be up to date with what is trendy and chooses her content mainly based of personal recommendations.

Stacy is aware of the dangers of her data potentially being leaked on the internet or utilized for the wrong purpose. She would have these conversations with her peers and professors quite often while she was in school. She tries her best to protect her personal information while using apps and services but understands that there’s only so much one can do.

Edward

Age: 32

Occupation: Product Designer

Edward lives with his partner working as a product designer mostly from the comfort of his home. He spends most of his day tending to their two cats and fulfilling his role as a product designer. The pandemic has really made him into a homebody.

When it comes to Netflix, Edward has been subscribing for almost 10 years now but he only has time to watch on average 2-3 hours a week. He sources his content mainly from Youtube trailers and other social media platforms. His friends are also busy working professionals and don’t have time to catch each other up on what shows and movies they’re watching, like they used to in college.

Edward thinks that if you’re online, you are inevitably going to be sharing your information with not only the app you’re using, but every other app out there. Your info will make its way around somehow. When it comes to digital privacy, he believes there is no such thing. At the end of the day, if there is a data leak, he just hopes that he isn’t the only one affected.

Why does this matter?

In the short run, users will suffer less by spending more time doing content research or reaching out to friends for recommendations. However these two methods: spending time on content research and reaching out social connection, are not sustainable in the long run. Users' age, employment and social status will greatly lower their chances of acquiring new contents from personal connections. At the same time, users' content research time greatly depends on how much time they have left outside study, work and family.

In the long run, this continuing dissatisfaction and disappointment could drive users away from Netflix. In regions where Netflix has not monopolized the entertainment market, users might ultimately migrate to other platforms, because they are still able to get access to the same amount of quality content but with less financial and data cost.

In order for Netflix to retain its existing consumer base, we recommend:

1. The recommendation algorithm should diversify its data sources and implement changes to its model to adapt to users’ evolving preferences and create more suggestions outside of users’ usual watch categories.

2. Every title should come with a short explanation of why it is being recommended and how the user’s data is being utilized. There should also be metrics to backup reasoning behind the rating percentage.

3. The system should give more weight to the like/dislike button and create incentives for more active user feedback. Moreover, the platform should inform users when their feedback is translated into recommendations.