This is Personal Playlist
The new Playlist feature for Sky Glass
At Sky I led the end-to-end product design delivery of the Personal Playlist feature for the Sky Glass TV. This case study details the process my team and I went through to deliver a feature that is currently enabling hundreds of thousand households in the UK to tailor their viewing experience to individual preferences.
Sky
Client
6 months, London 2023
Timeline
2 Designers
1 Researcher
1 Copy
1 PO
3 Developers
Team
Senior UX Designer
My role
Sky Glass was launched in 2021 - the first TV produced by Sky with a completely new streaming platform which offers users a fully aggregated and immersive experience by making consuming content easier than ever.
The problem:
Because Sky Glass is a shared TV platform, and TV viewing is often a communal household activity, segmenting the experience by profile has always been a point of contention. However, the lack of individual personalisation has increasingly frustrated users. Our challenge was to explore ways to personalise the viewing experience without relying on a traditional upfront profile selection system.
Playlist is a key feature on Sky Glass, allowing users to easily save content for later viewing by pressing the "+" button on their remote. Saved content is organised by category (Movies, TV Shows, Sports, Kids, etc.) and ordered by recency to facilitate discovery.
Insights:
User research confirmed the desire for personalised experiences on Sky TVs, particularly for content recommendations and easier content selection. However, users also valued the shared viewing experience. A key pain point emerged around Playlist: users expressed frustration with other household members' content cluttering their own lists. This insight led us to explore ways we could simplify content discovery.
Initial explorations and challenges:
We faced a significant challenge: simplifying content discovery without user profiles, a prerequisite for more targeted recommendations. Acknowledging the limitations of personalisation without viewer identification, we shifted our focus to customisation. Our strategy was to empower users to shape their Playlist experience based on established household viewing habits and routines.
High-level hypothesis:
We sought to empower users to personalise their Playlist experience and quickly locate saved items. To achieve this, we formulated a series of high-level hypotheses that informed the ideation process for the new Playlist feature.
Folders Vs Filters:
With the potential for user-created spaces within Playlist, we explored the optimal mechanism for their implementation. Two key approaches emerged:
-
Independent Folders: Folders for each household member, enabling users to specify content ownership, in addition to a shared folder for communal viewing.
-
Filtering System: A system allowing content to be assigned to multiple users, facilitating shared viewing while eliminating the need for a separate shared space.
Creating a list:
It was hypothesised that a familiar user interface pattern, analogous to profile creation but confined to the Playlist feature, would facilitate a more intuitive and personalised experience for creating new lists. Users are able to create multiple personal lists within Playlist by providing a name and selecting an avatar. This design was informed by a combination of exploratory research and competitive analysis, and subsequently subjected to prototyping and user testing.
Competitors analysis:
Although our competitors primarily utilize profiles, alternative approaches exist. Platforms like Instagram and Pinterest employ saving or tagging features, enabling users to create personalised folders while maintaining a comprehensive "All" section. We evaluated whether this pattern, common in single-user experiences, could be adapted to our shared viewing context for Playlist.
Design decisions:
Stakeholder discussions and evaluations led us to select the folder-based implementation (pending usability testing). Our rationale was that a mandatory "All" area would not sufficiently differentiate the new Playlist from the existing one and might be unnecessary for some users. An optional "Shared" folder, however, offered a more intentional approach to shared viewing.
Rapid prototyping facilitated stakeholder discussions and provided a clearer understanding of each option, streamlining the decision-making process.
Users trial:
The Personal Playlist feature underwent a three-month trial period with select users before its official launch. A subsequent online survey indicated strong positive feedback, with 75% of trial participants reporting an enhanced discovery experience. The survey also highlighted some opportunities for refinement in areas such as content removal and list editing.
Mitigating the "friction":
After wireframing and prototyping various add/remove options for Playlist, we selected a design that prioritized speed, visual impact, and a smooth, engaging user experience. Our hypothesis was that the benefits of the feature would outweigh any perceived friction from having to select a playlist.
A preliminary round of guerrilla testing enabled us to refine the prototype and resolve outstanding usability and UI issues prior to a more in-depth user testing study.
High-fidelity prototype:
After exploring various UI options, we created a high-fidelity prototype in Figma to represent the full user experience. This prototype, intended for usability testing, combined existing UI components with newly designed elements: an overlay for adding content to lists, circular tiles with colourful avatars for improved visual distinction, and new icons, including ticks for selection feedback and a "Shared" icon.
More testing:
We conducted moderated in person usability testing with 12 active Playlist users from diverse household structures (6 families, 5 single households, and 1 flat share). Our objectives were to evaluate feature understanding, list creation and editing, content management, and overall experience. Results were positive: all participants found the journey straightforward, saw value in the feature, successfully created, edited, and deleted lists, managed content addition and removal, and recognised the updated signposting. While initially associating the feature with "Profiles," users readily understood the distinctions and limitations from classic profiles; ultimately preferring names like "List," "Personal List," and "Person."
Liked what you've seen?
Let's work together,
say hi! 👋
Product launch:
The Personal Playlist feature, after incorporating minor improvements, was launched in May 2023. Data collected on December 3, 2024, shows over 272,000 active Personal Playlists in the UK, with an average of 2,960 new playlists created each week.
Adding / Removing behaviour:
The established method for adding content to the Playlist involved a single button press ("+") on the remote. The implementation of Personal Playlists (if users have created one) introduced a requirement for users to designate a specific playlist for content, thereby adding steps to a previously streamlined process. The challenge was to mitigate any potential friction resulting from this change.
