Apple Community Thread Recommendations
Design DRI
Mukund Asagodu Sanjeev
Design Team
Creative Director
Art Director
Head of Design (Manager)
Design Systems Manager
Prototyper
Design Producer
Partners and Stakeholders
Product Owner
Community Managers
Data Scientist (ML)
Business Intelligence (BI)
Front-End Development (FED)
Information Systems and Technology (IS&T)
Localization
Quality Assurance (QA)
Given Brief
Improve the community visitor experience by offering an easy way to find relevant community solutions - that they may not have found on their own via search engines.
Problems
Thread Relevance and Bounce Rate. Thread relevance is crucial as 95% of Community visits occur on thread pages. A mismatch between landing threads and customer problems leads to a high bounce rate of 74%.
Underutilization of Search. Despite the potential for self-solving with existing Community answers, less than 1% of users search during visits, indicating an underutilization of the search function.
Challenges with Community Participation. The requirement to register and wait for responses when asking the Community presents hurdles. New users might not realize the need to return for answers.
Duplicate Content Impact. Duplicate content hinders effective answer discovery, demotivates helpers, and contributes to unanswered questions for Service Delivery.
Business Goals
Help customers visiting Community thread pages, self solve through existing Community content, reducing the need to search again or create new posts.
Improve the community visitor experience
Offer users searching for information an easy way to find relevant, useful answers — without searching again — and reduce time to resolution.
Reduce duplicate questions
Make it easier to find answers in context, fewer users should need to post questions that have already been asked and answered by others.
Design Goals
Enhance user engagement by promoting relevant content discovery, and providing a seamless and unobtrusive user experience.
Challenges
Lack of understanding of the new model and its accuracy.
Existence of assumptions and obstacles about customer journey and design standards among team members.
Ensuring that the recommendations are relevant based on the thread's context and striking the right balance between avoiding overpromising and delivering meaningful recommendations.
Integrating recommendations within the thread while ensuring coherence and seamless interface, particularly in avoiding disruption to the flow of primary content.
Designing recommendations to be effective and non-intrusive specially on mobile devices with limited screen space.
Gathering feedback effectively for validating and also improving the ML model over time without disrupting the user flow.
Insights into what happened
Pivotal moments and progressions that unfolded during and after the design journey.
Inspired curiosity for clarity
Integrated designs seamlessly
Provided relevance and value
Preserved visual harmony
Proposed scalable designs
Inspired curiosity for clarity
Rephrased assumptions and obstacles into questions which created a subtle shift from uncertainty to curiosity among the team.
Guidance and clarity regarding Design requirements, including responsiveness and importance of grid.
Collaboratively generated a collection of big ideas.
Integrated designs seamlessly
Integrated recommendations subtly within the thread's context to avoid interruptions, and maintain reading continuity for both visitors and helpers by positioning them during natural thread breaks.
Offered a limited number of recommendations to prevent overwhelming users.
Provided relevance and value
Provided concise context and previews to aid understanding.
Achieved a balance between presenting relevant recommendations and incorporating minimalistic amount of cues to steer user engagement.
Preserved visual harmony
Preserved visual consistency and aligned with Design Standards to create a new pattern that resonates to Apple aesthetics.
Proposed scalable designs
Structured the design modularly, enabling easy updates without overhauls, aligning with an evolving ML model.
Designed adaptable interfaces to accommodate content variability as the ML model evolved in the future.
Collaborated to integrate a feedback process and analytics to comprehend model performance and user behavior.
Results
Enhanced user engagement through the facilitation of relevant content discovery, ensuring a seamless and non-intrusive user experience.
Cohesive experience that champion simplicity, accessibility, localization, and thoughtful consistency.
Compelling evidence with increased number of page views and decreased number of threads created per visits showed that visitors are more engaged in self-solving before seeking answers, highlighting the significant impact of recommendations on user engagement with Community content.
Consistent performance in FR and EN Communities week over week, period over period, which indicates the feature is stable and the FR launch using Machine Translation (MT) was a repeat success.
Expanding to remaining regional Communities to replicate the success.