A cover-first book discovery experience to recreate the feeling of browsing a physical bookshop online.
Role
Product Designer + Owner
I created the idea, planned the structure of the site and designed the site before directing Claude Code to build it.
Timeline
June 2026 - August 2026
Industry
Books
Tools used
The Problem
I felt book discovery in the digital space was too heavily based on data, tags and lists. Book cover design is such a big influencer in the decision-making process for readers in physical bookshops, so I wanted to create a site that did that in the digital space.
Goals
As this was a personal project, the goals are less about metrics and more about functionality and milestones.
Canvas
Must have a working canvas, letting users scroll through thousands of books
Variety
Have an extensive catalogue of books of all genres
Filter
Users can filter by mood, subjects, themes and pace as well as genre
Fun
It should be a product that I myself will use. If I wouldn't use it, why should anyone else?
Discovery
The largest section of research done was around competitors. Who were the big players, what was their key selling point, how did they showcase books, what did they tag them with for metadata, was there a site like my vision out there already. I discovered more sites than I was expecting to in the book space, many for tracking books, some for annotating and sharing with friends, but none of them were trying to rework book discovery the way I was.
Design
Canvas
The main attraction to the site, an infinite canvas full of books to scroll around and browse, all loaded near-instantly. One of the fundamental aspects of this design being successful is the cover quality, each cover has to be as modern and eye-catching as possible. A canvas full of old, poor quality covers would not achieve the same desired effect.
Search
I wanted the search experience to feel embedded in the canvas, keeping you on the same page and have books magically appear as you type.
Shelves and sharing
Allowing users to display their books in a cover-first way was important to the consistency of the site experience. It lets people appreciate and showcase the books in their shelves in a much nicer way than competitors like Goodreads do.
Users can also export their shelves into an image in a few simple clicks, allowing them to share them on their socials, key for marketing and encouraging word of mouth.
Details Modal and Page
I wanted the browsing experience to be quick and seamless without long page load times every time a user wanted to learn more about a book. Because of this I used modals that appear on cover-click that provide all the relevant details of the book, opening instantly and keeping the user's coordinates on the canvas for when they close the modal and want to continue where they left off.
The SEO downside to modals is that the information inside them can't be crawled by search engines so we lose all the great metadata inside them like the description, Turner's Opinion and the various tags. Therefore I decided that every book also requires a dedicated page as well as a modal, almost entirely for page search indexing.
Filtering
Books are more than a genre, so I tagged all books with various categories to allow for more specific filtering. These can be used in an AND or OR filtering method.
Mobile
With significantly less screen real estate, I had to completely rethink how the site would work on mobile. I opted for a more random/filtered suggestion feed that users could narrow down by tag and then be served books on an endless scrolling feed. This is a common behaviour now with social media sites but I think it still maintains the cover-first experience that's at the heart of the site.
What went live
35,000+ cleaned, categorised and tagged books
Desktop canvas loading 2,000 books at a time
Responsive mobile discovery feed
Search and multi-dimensional filtering
Custom shelves and social-image export
Modal and indexable page for every book
95k+ pages, for books, authors, series, tags and similar books
Admin portal for catalogue management
Hurdles faced
Book data
There is not one single source of truth available for book data. There are several partially complete databases though, so I had to merge these sources and use AI to match and clean them, creating my own database of 700,000+ datapoints for every book as well as having a high-quality cover for each.
Canvas performance
Loading 35k books on a canvas at one time proved to be a huge technical obstacle that became too big to overcome. I had to pivot and set a 2k limit on books loaded on the canvas at one time. Then I added a "load new books" button when a user has seen all the current books.
Cover quality
The covers I was able to get via the free-to-use APIs like Google Books and Open Library were poor quality. To get round this I had to use other methods of obtaining new, high-quality covers.
Next steps
Now the site is live, I am working on more features that users can use to find their next book, such as making Turner, the AI librarian, an interface that users can actually talk to to get incredibly specific book recommendations.
