Personalized Book Discovery
What?
A personalized book recommendation feature concept for Goodreads focused on improving user engagement and retention.
For who?
Casual and avid readers who want more relevant, emotionally aligned book recommendations.
My role
End-to-end UX design, including competitive analysis, user research, synthesis, wireframing, usability testing, and high-fidelity prototyping.
Project duration
2 weeks
Tools
Figma
Why This Matters
Goodreads has remained one of the most popular social sites for readers, with estimates of over 150 million users as of 2024. However, after Amazon's acquisition of Goodreads in 2013, the site has continued to receive criticism over its outdated interface and many users find themselves migrating to other sites like StoryGraph and apps like Fable.
Project Goals
Understand what causes migration from Goodreads
Understand user pain points with Goodreads' current interface/experience
Help Goodreads users feel more in tune with the platform and its reading community
The Solution
I designed a personalized recommendation experience that helps readers discovery through mood-, genre-, and theme-based suggestions tailored to their reading history and preferences.
How I did it
The Competition
I analyzed 3 Goodreads competitors to understand how they differ and what offerings make them appeal to readers.
Core Insights
Theme 1: Personalization appeals to users and can enrich their reading experience.
Theme 2: Competitors offer curated insights and metadata, which is something that Goodreads lacks in depth.
Theme 3: Community engagement remains a common feature among Goodreads and its competitors.
Learning From Readers
5 participants
I recruited 5 interview participants (ranging from new readers to avid readers and Goodreads migrators) to discuss their engagement with Goodreads and other book tracking platforms.
13-16 questions
My interview guide had 16 total questions, though only 13 were used in some testing sessions depending on the participants' experience with Goodreads or alternative platforms.
8 insights
After compiling my interview findings into an affinity map, I came up with 8 major insight groups that revealed pain points and areas of opportunity.
Interview insights
The most common issue among users was the lack of personalization in Goodreads' book recommendations, validating my findings from my competitive analysis regarding the emergence of personalized metrics in other platforms.
User-Informed Personas
Defining the Problem
Problem Statement
White Goodreads has remained a staple book tracking platform for both casual and avid readers, users find frustration in its discovery features. Goodreads' recommendations often fail to match users' reading preferences and history, leaving new readers frustrated and motivating avid readers to migrate to other platforms.
How might we…
improve the discovery experience to provide recommendations that better match users' preferences and reading histories, thus supporting the growth of their reading habits?
Designing for Personalization and Retention
Identifying essential features
I conducted an audit of Goodreads to identify its most essential features and ideate improvements that might be "surprising and delightful" additions to the user experience. Based on the problem statement and insights from users, I decided on a "Your Next Read" feature that would provide tailored suggestions based on user feedback about their current read.
Defining user flows
I mapped out my idea with a user flow chart to detail the screens I need to design and understand how the user would interact with this feature. I also used this user flow to design 7 low-fidelity screens based on Goodreads' existing UI patterns, ensuring they were grounded in feasibility and familiarity.
Validating Designs
2 goals
The purpose of this low-fidelity testing session was to validate the feature's usefulness for Goodreads users and refine the intake criteria for book suggestions.
3 participants
I conducted this first round of tests with 3 participants from my user interviews.
3 tasks
Users were given the scenario of having finished a book, wanting to update their progress, and find a new book to read. Each task supported one of these three steps/goals.
Testing Summary
Finding 1: Participants experienced a disconnect between their input criteria and the suggested results, revealing a need for clearer language and presentation.
Finding 2: All participants suggested new feedback chips to better describe what they look for in a new book.
Finding 3: Participants expressed that having simple choices made the experience very easy to use, showing the value of simplicity in reducing choice fatigue and retaining engagement.
Priority Revisions and Increasing Fidelity
High-fidelity wireframes
Applying Goodreads' brand, typography, and visual styles evolved the low-fidelity wireframes into high-fidelity wireframes ready for prototyping and testing. These screens were iterated based on user feedback, focused on reducing cognitive load and improving clarity.
A/B/C screens
In preparation for high-fidelity testing, I created 4 screen variations of the results page to support A/B testing:
Variant A: Original screen with unsorted list of recommended titles and no descriptions.
Variant B: Similar to Variant A, but with descriptions to accompany each title.
Variant C: Results sorted by the user's input criteria (e.g., similar genre or similar plot).
Second Round Design Validation
5 participants
I retested the prototype with 3 participants from the lo-fi testing sessions and 2 new participants who have not seen the previous wireframes.
2 testing phases
The first phase of the test was a moderated usability test similar to the lo-fi testing. The second phase was an A/B testing session for the results page.
Testing Summary
Finding 1: New participants enjoyed the suggestion experience and expressed that it was straightforward and clear, reaffirming my findings from the low-fidelity testing sessions.
Finding 2: Returning participants enjoyed the addition of the book synopsis on the "Suggested Titles" page.
Finding 3: Participants liked Variants B and C from the A/B testing session. Therefore, a combination of the two screens would provide the most clarity
The Final Design
Putting It All Together with High-Fidelity Prototyping
The final prototype reimagines book discovery on Goodreads through a more personalized and engaging recommendation experience. Instead of relying on popularity-driven suggestions, the experience allows users to discover books based on mood, themes, genres, and more, curating recommendations that feel more relevant to what readers actually want in the moment.
Write a review
Similar to Goodreads' existing flow, users will be prompted to rate their book and write a review. Once they've completed this step they'll be prompted to either finish or find their next read.
Describe your last read
Users will be asked whether they want a book that is similar to one they just read or a new book. If they choose a similar book, they will be asked to describe what they liked so the algorithm can find books that match those criteria.
Start reading
The user's final step is to choose their next read from the list of suggested titles. Users can view each book's rating, synopsis, and similarities to their previous read. From this list, users can then access each book's full page and add it to their list.
Reflection
This project taught me that personalization is about more than just algorithms, it's about helping users feel understood in moments of decision-making. Designing for both casual and avid readers also showed me how different users approach reading, from seeking inspiration and guidance to wanting precision, analytics, and control.
This project also strengthened my understanding of iterative design within an existing ecosystem. Testing revealed that even valuable features can feel overwhelming without clear task flows and focused interactions, reinforcing the importance of simplicity and clarity in user experiences. Overall, the project deepened my appreciation for grounding design decisions in real user behaviors and designing systems that encourage long-term engagement rather than one-time interactions.
