Picsart content browsing on inspiring feed
In Picsart, both prosumers and consumers face content discovery challenges. To tackle this and reflect user diversity and simplify conent discovery, we're comprehensively redesigning the Browse page, focusing on enhancing merchandising and recommendation features for a personalized editing experience. Leveraging user segmentation and behavior analysis, our aim is to deliver tailored recommendations, driving engagement and loyalty.
🤔 Problem:
Enhancing the browsing experience is crucial to meet the rising demands of Picsart's professional user base. Anticipating a surge in Prosumer and Business users, urgent action is needed to overcome current obstacles:
- Relevance Challenge: The Browse page falls short in meeting professional users' needs, risking alienation.
- Tone Mismatch: There is a significant risk of presenting content that leans more towards fandom-oriented material rather than meeting the standards expected by professional users, potentially leading to dissatisfaction and disengagement.
- Past Strategies: Previous approaches like gross actions for recommendations and content separation into single asset type feed within separate tabs have proven insufficient and outdated for evolving professional needs.
Furthermore, specific issues exacerbate user experience:
- Content Clarity: Users struggle to differentiate between various content types, resulting in mismatches between their expectations and the content presented to them.
- Personalization Gap: There is a notable absence of tailored content catering to users with diverse and unshared interests, leading to a lack of engagement among certain user segments.
- Relevance for Advanced Users: The current content offerings fall short in satisfying the requirements of prosumers and marketers, posing a risk of alienating these advanced users.
- Diversity & Freshness: There is a pressing need to enhance content diversity to prevent user disengagement and maintain interest over time.
- Feature Familiarity: New users often struggle with understanding and utilizing key features such as remix, replay, and regenerate, hindering their overall experience and potentially leading to frustration.
🔎 Research:
When Tackling our multiple layer problem, we addressed the following aspects:
• Interactions: We prioritize interactions that enhance learnability and engagement with the content.
• Structure/Layout/IA: Developing a structure that aids understanding of content origin is crucial.
• Algorithmic Logic: Ensuring that the underlying algorithmic logic maintains relevancy is paramount.
To address these challenges effectively, we conducted rapid cycles of research aimed at exploring and evaluating three key design principles: valuable, usable, informative and personalised. While the problems are evident, our focus is on determining the most effective design and content solutions. Through iterative research and design cycles, our objectives were to:
• Understand the browse page's value from the user's perspective, identifying what attracts and retains their interest.
• Utilize layout and interaction design to enhance usability and drive engagement.
• Improve the learnability of essential concepts such as remix and replay to encourage increased engagement.
• Identify how personalised should the browse feed be.
Driving User Behavior: Users are discovering and engaging with a broader range of content, tools, and topics, marking progress from seeing content to learning and trying it out, ultimately leading to visits to activity conversion.
Design & Research Principles: driving visits and engagement, the feed must be valuable, usable, informative and personalized.
What We Knew
• Users struggle to grasp the relevance of the content they encounter.
• New users find the purpose of the discovery feed unclear.
• Concepts like Remix, Replay, and Regenerate are sources of confusion for users.
What We Didn't Know
• Unveiling the potential value of the browse feed from the user's perspective and identifying factors that will maintain their interest.
• Understanding how the layout and design of the feed can create momentum and drive user engagement.
• Finding effective strategies to address the persistent challenge of learnability in a manner that resonates with users.
Considerations
• What do users perceive as the value of the feed, and what factors influence their continued usage and return visits?
• What feed layout and interaction design do users find most intuitive for easy navigation, discovering a variety of content, and engaging with it?
• What are the key factors affecting the learnability of our concepts, and how can we enhance this for users?
Overview of Research Cycles: We conducted rapid cycles of design and research to explore and evaluate our design principles' effectiveness and their impact on the behavior changes outlined in our KRs.
Cycle 1: Value
Aim: Understand how to drive engagement based on what the user finds valuable.
Approach: Unmoderated interviews.
Outcome: Understanding of value and recommendations for content and categories.
Cycle 2: Layout
Aim: Identify what makes a feed usable and useful for the user.
Approach: Unmoderated testing.
Outcome: Recommendations for layout design and improvements to feed interactions.
Cycle 3: Personalization
Aim: Understand the level of personalisation users need when interacting with the feed.
Approach: Moderated interviews.
Outcome: Recommendations for personalization on the feed.
Cycle 4: Learnability (tbd)
Aim: Understand what makes a good learning experience and evaluate learnability concepts.
Approach: Moderated interviews.
Outcome: Recommendations for improvements to learnability.
Final Evaluation
Aim: Evaluate the design principles of the feed and their impact on the behaviours we are driving.
Approach: Unmoderated user testing.
Outcome: Core concepts for A/B testing.
Cycle 1: Value
After conducting unmoderated interviews aimed at understanding how to drive engagement based on what users find valuable, the research yielded valuable insights and actionable recommendations for content and categories. Specifically, the outcomes include:
• Enhanced Understanding of User Preferences: The research provided a deeper understanding of what users perceive as valuable content on the platform. By analyzing user feedback and preferences, we gained insights into the types of content that resonate most with our audience.
• Identification of High-Value Content Categories: Through the interviews, we identified specific content categories that users find particularly valuable. These insights enable us to prioritize and curate content effectively to meet user expectations.
• Recommendations for Content Strategy: Based on the findings, actionable recommendations were developed to refine our content strategy. These recommendations include suggestions for creating and promoting content that aligns closely with user interests and preferences.
• Improved Engagement Strategies: The research outcomes also informed the development of engagement strategies tailored to enhance user interaction with valuable content. These strategies aim to increase user engagement and retention by delivering content that adds significant value to their experience.
Overall, the research outcomes from Cycle 1 provide valuable insights and actionable recommendations to drive engagement by delivering content that users find genuinely valuable and relevant.
Cycle 2: Layout
Cycle 2 was centered on determining the most user-friendly feed layout. Through testing three distinct layouts with six users - comprising three solopreneurs and three consumers - we derived valuable insights. Based on the results, we advocate for adopting a one-column feed format with two tabs. Additionally, our recommendations include reducing the frequency of carousels, implementing a multi-action CTA, and incorporating a blend of creative actions and save options. It's important to note that our primary objective is to ensure that any adjustments made do not compromise the user experience, rather than specifically generating insights tailored to this user segment.
(Open the prototypes in a new tab or play the video)
👀 Prototype 1 - two column feed with two tabs
👀 Prototype 2 - one column feed with 2 tabs
👀 Prototype 3 - One column feed with one tab
Preference for Two Tabs:
• Majority of users (5 out of 6) preferred having two tabs, allowing them to view different types of content at different times.
• Consumers typically divided their browsing and social activities, while solopreneurs appreciated the efficiency of having two tabs. Recommendation: Two tabs
Advantages of One-Column Feed:
• Research indicates that using carousels affects how users perceive the feed. While a two-column feed shows more content, it can feel crowded when combined with carousels.
• Users preferred a one-column feed as it allows them to focus on one large piece of content at a time without distractions. Recommendation: One-column feed
Addressing Carousel Frequency:
• Users appreciated the variety of content in carousels, ranging from creators to hashtags and Spaces (communities), as it helped them find inspiration.
• However, particularly solopreneurs found that the carousels appeared too frequently and disrupted the feed's flow.
• It's crucial to carefully consider what content to place in the main feed versus the carousels. Recommendation: Reduce frequency of carousels
Prioritizing Creative Actions:
• Users mainly saved content during browsing but leaned more towards trying or editing when using prototypes.
• This suggests that users recognize Picsart as a creative app where they can do more than just save. Creative actions are as important as saving on the browse page.
• However, some users were confused about the meaning of "Remix." Recommendation: Mix of creative & save actions
Benefits of Multi-Action CTA:
• Users favored the cards in the two-column feed because they focused on the images and were less busy and distracting than the one-column feed. Recommendation: Multi-action CTA (further details to be discussed in the next case study)
Cycle 3: Personalization
As we embark on the development of the new browse feed, a central obstacle we face is ensuring content relevancy. One pressing question on agenda is: How personalized should the browse feed be? To swiftly gather preliminary insights, we tapped into users within Spaces (community). We tasked them with creating a collage to depict their ideal inspiring feed and engaged in discussions via comments. Our findings revealed that the interviewed users aspire to a diverse and dynamic discovery feed that strikes a balance between catering to individual interests and exposing them to new content and artists. For this specific user group, fostering a sense of community and appreciation for all forms of art is paramount for an engaging and satisfying discovery experience. Building upon the technical investigations conducted by the discovery team, our next phase involves testing prototypes of personalised feeds with broader user segments.
“Art that inspires me to do better.”
“Everyone should feel equally featured and appreciated.”
“Expose us to unexpected and refreshing ideas, breaking the monotony and encouraging exploration.”
Level of Personalisation Desired
User-Specific Interests and Inspirations: Users express a strong desire to see content that aligns closely with their individual interests. Interests can mean both content medium (such as Gen AI and old school edits) or art theme (such as surrealism or dark art).
“I would like to see: "traditional" drawings and edits (i.e. not generated by AI), because these are contents in line with my interests.”
Variety vs. Niche Content: While some users seek content that matches their specific interests, there is a notable demand for variety, indicating that users appreciate a mix that includes but is not limited to their usual preferences. This suggests a balanced approach to personalization, where the feed introduces users to new content types and artists outside their regular consumption patterns.
“We like the variety of artistic genres, and not vain repetitions of the same thing.”
Needs & Wants
Separation of Content Types: Users want clear segregation of different content types, such as tutorials (replays), challenges, and AI-generated content from the main feed. It is possible that their preferences are led by the tab approach already taken. This means that organisation of the new discovery feed will be an important aspect of navigation.
“I want to see all of the art forms except replays. Replays should stay out of the main feed.”
Freshness and Relevance: There is a clear frustration with seeing outdated or repetitive content. Users expect the discovery feed to feature recent and relevant content, highlighting a need for a dynamic and constantly updated feed.
“This is what keeps popping up every time 🤷🏼, actually that boy is in one of the top spots since weeks. So, no wonder, I never even start scrolling.”
Inclusion and Exposure: Users appreciate an "all-inclusive" approach where various art forms and artists (regardless of their following size) are featured, ensuring exposure and appreciation for a wide range of creativity. This also includes a desire for features like an artist spotlight to encourage and highlight exceptional work.
“I want an artist spotlight corner in that feed too, like we had earlier.”
Type of Content Desired
Diverse Art Forms: There's a strong desire for a wide range of content, including old school edits, photography, drawing, AI-generated content, and specific themes like space, horror art, and romance. This diversity reflects the community's varied interests and the need for a feed that caters to a broad spectrum of artistic expression.
“I would like to see a variety of content: ‘old school editing, photography, drawing, AI generated.”
Educational vs. Inspirational Content: While some users value tutorials and learning content (replays), there's a consensus that such content should be optional or housed in a separate section, allowing the main feed to focus on inspiration and discovery.
“Maybe the replays should be shifted to a separate tutorial tab 🤷♀️ I actually just post them, because they explain how I work.”
Community and Creativity: Users express a need for content that fosters community interaction, creativity, and personal growth. They seek inspiration through a discovery feed that challenges them, exposes them to new ideas, and reflects a wide range of cultural and artistic influences.
“Picsart has always been an ‘All Inclusive’ app. Everyone should feel empowered to express themselves with their art and feel equally featured and appreciated.”
Special thanks to Dr Charlotte Pyatt-Downes, for being an amazing research partner in conducting this research cycles and analysis results.
🚀 Final designs
Following the insights gathered from our extensive research efforts, we have implemented significant changes to enhance user experience on the platform:
Tab Navigation: We have introduced two main tabs - "Inspiring" and "Following" - to streamline content discovery. The "Inspiring" tab offers curated content tailored to user preferences and interests, while the "Following" tab displays content from creators users are following, as well as relevant communities and hashtags.
Layout Optimization: After careful consideration, we have adopted a one-column feed layout to provide users with a more immersive browsing experience. This layout choice allows content to breathe and prevents overcrowding, ensuring users can focus on one piece of content at a time. Additionally, we have strategically placed four content pieces between each horizontal carousel to maintain user engagement and flow.
Content Tagging: Each card within the feed now features tags such as "popular," "suggested," and "for you," providing users with insights into the content type and relevance. This tagging system aims to improve content discoverability and user understanding.
Personalized Carousels: Carousels within the feed now dynamically match users' preferences gathered from onboarding and past interactions. Users will encounter suggested templates, communities, tutorials, creators, and more, tailored to their individual interests, enhancing their browsing experience.
Multi-Action Call-to-Action (CTA): We have retained the multi-action CTA, offering users a variety of creative actions directly from the feed. This feature empowers users to engage with content in diverse ways, further enriching their creative journey on the platform.
These changes reflect our commitment to continuously improve and innovate, ensuring that users can discover, engage with, and create content in a seamless and personalized manner.
🏆 Final prototype
🧠 Hypothesis
We believe that enhancing the Browse page in Picsart with better content differentiation, streamlined user flow, and personalized recommendations for our diverse user base including first-time users, prosumers, and marketers, content tailored by user segmentation, will achieve higher user engagement, increased satisfaction with content relevance, and greater utilization of platform features like regenerations, remix, and replay.
Why is it valuable/ impactful
Aligning Browse with Prosumer/Business Growth
Empowering Prosumers: Focused on small and midsize businesses and marketers.
Key Objective: Deliver the right content and tools directly through the Browse screen.
Outcome: Significantly expedite campaign creation for these user segments.
If we are successful we will empower prosumers (small and midsize businesses and marketers) with the right content and tools that significantly expedite campaign creation right from the Browse screen.
Aiming for
Enhanced personalized discovery
Streamlined user onboarding and learning
Diverse and dynamic content interaction
How will be solve the problems we have: Leveraging Multiple Models for Enhanced Recommendations
Multi-Model Capability: Ability to create, manage and iterate on various models for nuanced user understanding.
User Segmented Onboarding Data: Utilizing insights from tailored onboarding experiences to refine recommendations.
Prosumer Spaces Insights: Leveraging data from related user interactions to cater to specific segment interests.
🔎 Market research
Conducting market research, we analyzed applications like Facebook, Twitter, Instagram, LinkedIn, and others, as well as music and education platforms like Headway, Canva, Creative Cloud, TED, MasterClass, and Netflix. Common design elements emerged, featuring either a content feed or a horizontal shelf of algorithm-based recommendations.
Our findings highlight key details, including:
The importance of user-friendly text interactions
Incorporating subtitles or descriptive lines enhances learnability
Providing users with hints about their location and activities
Facebook ranking signals: Friends and followers, engagement level, content type, and content quality.
Instagram ranking signals: activity, interaction history, post information and content creator details.
Pinterest ranking signals: focuses on website ownership, quality and engagement levels.
Twitter ranking signals: user interaction, popularity and recency, relevance, diversity and media files.
Netflix ranking signals: views, rating, viewing history, preferences, watchlist activity, new.
LinkedIn ranking signals: engagement and interaction, relevance, recency, connections, content quality, hashtags.
Data evidence
Based on data evidence, it's observed that a significant portion of users follow at least one user, yet only a small fraction click on the following tab. This suggests a discrepancy between user behavior and platform features, indicating an opportunity to enhance the visibility and utility of the following tab.
Furthermore, analysis reveals that a considerable percentage of users frequently switch between home tabs. Additionally, a noteworthy proportion of users engage in extended browsing sessions, with many continuing scrolling beyond a certain threshold. This data underscores the importance of optimizing content presentation and navigation to accommodate user browsing behaviors effectively.
Moreover, a substantial increase in T2I (Time-to-Image) open was observed following the transition from a two-column to a single-column feed layout, with only a minimal decrease in Editor Open. This indicates that the redesigned single-column feed successfully encouraged user interaction while maintaining essential user actions. This highlights the potential for layout optimization to positively impact user engagement metrics.
Relevancy
With our recent investment in customer segmentation, we are dedicated to providing a truly personalized experience for our users. Our focus is on fostering deeper engagement and satisfaction through several key initiatives:
Firstly, we aim to implement multi-model recommendations, offering finely tuned suggestions tailored to each user segment. This approach ensures that users receive content that aligns closely with their preferences and interests.
Secondly, we are committed to dynamically responding to evolving user behavior patterns. By closely monitoring user interactions and feedback, we can adapt our recommendations in real-time to better meet their changing needs.
Additionally, we plan to introduce a recommendation engine customized for each item type, enabling more precise content delivery. This targeted approach ensures that users receive recommendations that are highly relevant to the specific types of content they are interested in.
Lastly, we intend to implement a recommendation engine that dynamically defines content type proportions. This allows us to optimize the balance of content types presented to users, ensuring a diverse and engaging experience.
💪 How will we measure success?
Objective: Users are discovering and engaging with a broader range of content, tools and topics.
KR 1: Engagement: Overall visit to action conversion rate. Prosumers see more relevant content in discovery feed and are more engaged with the content, i.e. perform more actions, like save, share, editor open, regenerate, etc.
Current value: x%
Target: x+4%
KR 2: Diversity: Prosumers interact with more diverse content, tools and topics in their feeds. In addition to replays, photos and stickers, users will be recommended with relevant Spaces and more Gen AI content.
Current: the percentage of actions on Gen AI content is x%.
Target: mix actions on Gen AI + Spaces to x+8% for Prosumers.
⚒️ Tools:
- Figma
- Looker
- Loom
- Miro
- Hotjar
