Pickflix
A movie recommendation system for groups
Contribution
User research,
Prototyping,
End-to-end design
Team
3 Developers,
1 Product Manager
What should we watch?
In 2017, 52.4% of the US population reported watching TV as a social activity with friends and family. However, groups often have trouble finding a movie to watch because balancing a myriad of preferences and interests is difficult and time-consuming. If they aren’t able to find a movie that works for everyone in the first 10 minutes, they settle for something not everyone enjoys as much, leading to lower engagement rates.
How might we help groups balance a myriad of movie preferences?
But, why are groups having trouble deciding what to watch?
To answer this question, I conducted structured user interviews, a guided group discussion, and literature reviews. From my research, I found 3 main pain points.
1. Mood
Someone’s “in-the-moment” preferences can be very different from their general likes.
2. Trust
People trust recommendations provided by their peers more than what a platform tells them.
3. Conformity
People often feel pressured to conform to group norms or opinions, especially when they don’t know other members as well.
Using research to reframe the problem
How might we reduce social conformity amongst groups and create a recommendation system that users can trust?
What's the value of solving this problem?
From my research, it became clear to me that the goal of my design shouldn’t be to improve the social movie-watching experience. Rather, I needed to develop a scalable framework that improves the decision-making experience. Doing so would provide my team the opportunity to scale this solution to other use-cases in the future, such as deciding which restaurant to go to or planning a group-vacation.
Turning pain points into design opportunities
Exploring the solution space
With my design opportunities in mind, I began to ideate potential solutions. I took this as an opportunity to explore the different media through which users may interact with such a system.
But first, a feasibility check
Of all the media I explored, a cross-platform desktop and mobile website appeared most scalable and adaptable. Above all, users were most likely to have access to either device when watching movies with others.
Initial Explorations and Testing
With the direction of creating a mobile-first website, I began sketching ideas. Once I had a basic system flow ready, we pilot tested our idea by sending out a link to a working website to over 200 potential users. For this test, we had two goals:
1. How do users feel about answering two preference questions during onboarding?
2. How do users interact with the recommendations displayed for their group?
Testing onboarding flow questions and interactions with recommendations
1. Smooth onboarding
Asking members their genre and streaming platform preferences was not intrusive and the context behind the questions was well understood.
2. Low engagement with recommendations
Most users did not click into any of the 9 movie recommendations displayed to their group.
"I liked how quick the onboarding was, but I don’t see anything here that I’m particularly interested in right now. I’ll try again later, maybe..."
Improving the recommendations experience
Collections instead of individual movies
Something I observed during our pilot test was that people needed just a single relevant recommendation to encourage them to engage with Pickflix’s content. So, I decided to test the idea of displaying collections of movies to represent a broader mood instead of a individual movies.
Inspiration: How can I capture a broader mood?
Sketches: How can I represent a collection of items?
Members can engage with individual movies within a collection
Moving away from scrolling
Instead of creating a primarily scroll based discovery experience, we created the option for users to generate a new set of recommendations at the hit of a button.
Design explorations for movie recommendation page
Creating a simple filter and card system for recommendations
Surfacing recommendation endorsements
People trust others’ opinions more than they trust an algorithm. To improve their faith in Pickflix’s recommendations, I decided to indicate which other members of a group have liked a particular movie on the main recommendations page. This information adds another layer of credibility to the movies Pickflix displays and further encourages users to engage with movie tiles.
How do we measure our success?
1. Number of successful groups
Successful groups are unique groups that have been able to find a movie to watch using Pickflix. With our current design, this action is measured by the number of users who click on the option to watch a movie and leave Pickflix.
2. Groups created with more than 1 person
This measure of success contributes directly to our goal of showing members how much they have in common, engage and inspire them to come together more often.
Next Steps
With these metrics in mind, my next step for this project is to test how users respond to the design changes made to the recommendations experience. These changes can be tested by A/B testing two versions of the website: the first where the recommendation screen displays a combination of individual movies and collections of movies. The second version will display only collections of movies. Measuring engagement rates of both version can help us understand if the decisions made are in the right direction.
What I've learnt
As a designer, I learned the importance of being cognizant the harmful potential of my designs. There were early iterations of the system that allowed users to maintain an anonymous identity, even with groups of close friends, to give them complete control and comfort. However, such a user state may have exacerbated a user's discomfort with expressing their own opinions and accepting those of others.Moving forward, I hope to iterate my designs to address this impact in a more effective way. Conducting observational studies of groups interacting with PickFlix, followed by interviewing each group member individually could yield potential insights on this issue.
Keep going!
Up next: Problem Roulette
An updated set of icons for Problem Roulette's design system
GO TO PROJECT
