Bookish
Streamlining the search and product discovery experience to build user trust and convert browsers into buyers.
The brief
Bookish is a Spanish book platform created in 2017. What started as a curated subscription box has grown into a full literary ecosystem—combining e-commerce, flagship bookstores, and Crush Fest, a major Young Adult festival.
Its differentiator has always been the same: human curation over mass catalogue.
We built on a shared foundation: identical book data, but distinct bookstore brands.
The data is pulled directly from DILVE, Spain’s official publishing database, using the industry-standard ONIX format
The
problem
How do we create a trustworthy book page that adapts seamlessly to every book in the catalogue, from the best-documented to the barely-documented?
The key insights
How readers chose a book
1
Theme and subject
mentioned by 83% of the readers
2
Author Reputation
mentioned by 52% of the readers
3
Personal Recommendations
mentioned by 41% of the readers
FGEE 2025, national data
The importance of recommendations
62%
of readers say they look for information or recommendations before choosing new reads by 83% of the readers
La Federación de Gremios de Editores in 2025
Bookstores and human recommendations were also highlighted during our research interviews with the following quotes
based on 5 user interviews
Written note from the bookseller
“The little handwritten card inside where the bookseller left a note... it'll really make me want to read it.”
Human recommendation from the bookseller
“I like having human contact in bookstores, a real person suggesting a read they personally enjoyed.”
Endcap that catches the attention
“I buy books displayed on the endcap (bestsellers usually) and the visual just catches my attention”
Ideation & Testing process
Using Bookish Design system and crafting interactions for features prioritization and testings
People also read recommendation
Mark as read option
Leave a review
All the reviews visible at first sight
Time reading
15
testers in total
5
participants per testing phases
5
testing rounds with 1 concept testing and 4 usability testings
1
Homepage & book page: information hierarchy
2
Book page hero section: A/B testing
3
Book page datasheet information from DILVE
4
All results page: Card sorting and A/B testings
5
Book page author and reviews sections: A/B testings
The design
The admin onboarding flow
We came-up with 4 admin profiles: parent, teacher, principal and administrator.
The goal was to showcase diversity and also ease the onboarding process for our users.
Parents or institutions can create an admin account to add a child or a list of children that will have access to the platform.
We decided to add a bulk subscription to avoid friction while connecting students to an admin account.
Gamification on the platform
As highlighted in the Design research, gamification and the possibility to touch and practice is really important in a child learning process.
We added these elements in the platform and tried to recreate as realistically as possible the virtual environment of a piano lesson.
3 All results page
We focused on the filters information hierarchy and the fact that animal salvaje is a book in multiple languages.
FILTERS
Card sorting helped define natural category groupings
A/B testing proved users preferred filters collapsed by default to prioritize scanning the book grid.
So we came up with Price first, Language second and Category third.
LANGUAGES
Testers wanted to select their language early on to avoid mistake, we added a dedicated space to the product card.
5 testers confirmed that a text format provided the clearest user experience.
4 The book page
We conducted a heuristic analysis and here are our findings.
HERO SECTION
The author and the translators are not differenciated clearly. As for click and collect we can't see in which library the book is available. In addition, the majority of users wanted a version where you could chose the book format inside the product page.
BOOK DETAILS
You can see lots of information so we wanted to ensure that people could understand clearly what the book is about, streamline redundant information like the critics in the sinopsis and make the "read more" buttons more visible.
We refined the spec sheet after user testing. We dropped duplicate info from the header (such as autors and ISBM), turned technical jargon into plain, simple spanish and we added 'Original Language' based on direct user feedback.
REVIEWS
Reviews are displayed in an unfiltered, scrollable list, making them difficult to navigate in case of many reviews.
AUTHOR SECTION
In the original design, the author's information sit at the same visual level as the purchase block. It is competing directly for the user's attention, also we can't see directly the book from the autor selection which is below.
Key learnings
Trust beats volume
Presenting the right info at the right moment builds confidence far better than overloading the page.
The data was already there
We didn't need more content. We just needed better design to unlock its value.
Designing for the hardest case made the easy better
Designing for the most complex cases ultimately made the experience better for everyone.
Next steps
Validate at scale
While user testing provided valuable insights, scaling the tests across the full audience is essential to validate our findings at scale.
Scale across the other brands
Test and apply the template to all four store brand identities.
Extend the review community
The review volume needs to grow to fully activate trust on the new functionality we designed.
