Goodreads Recommended Reviewers
Goodreads is a community for people to share their reading lives with one another. However, it’s hard to find people that share the same reading tastes. Helping new members quickly find people who reflect their own interests represented one of our biggest opportunities for new member engagement in 2018. I worked with my team to develop a solution that introduced new users to like-minded content creators during the account creation process.
Prior to this, all new Goodreads members were prompted to connect to Facebook upon creating an account. This simple call to action was enough to get some people to connect, but even in this best-case scenario the community simply wasn’t dense enough to guarantee a rich experience. The odds of you having a Facebook friend active on Goodreads are low enough— the odds of one reading books that you’re interested in are astronomical. As a result, most people weren’t having a social experience when they came to Goodreads.
Despite these headwinds, Goodreads has a rich community of avid readers writing high-quality, in-depth reviews of books in all genres on a daily basis. New users simply had no way to discover them.
In 2018 we developed a targeting service that would identify a set of ‘Top Reviewers’ based on that member’s favorite genres, and began exploring the best way to deliver the service to new members in our community.
My job as UX Designer was to determine the best way to deliver these recommendations to a new user. I conducted user research and concept tests, built wireframes and prototypes and iterated on a design before arriving at a final recommendation, for which I oversaw the implementation and release.
Our project began with survey-collection to determine what members were looking for in a reviewer or book recommendation, and identified key elements that would lead to a strong social match. The most important elements in building a successful connection were trust and similarity of taste. Based on that, I developed a series of wireframes that represented previews of different user profiles.
We displayed these concepts mocks to new users and members, asking them to judge each person based on similarity to their own reading tastes and their level of trust in their judgment.
I found that users most often gravitated towards users who had read books that they were familiar with and had read before or had positive impressions of. While participants most often trusted users with positive text reviews of books they had read before (top left), the same content was much less effective if the user was unfamiliar with the book. On the other hand, users were much more likely to recognize books if there were multiple covers shown (column 3). Based on these and other findings , I developed the following design.
This object view displays the most relevant information for a new user gauging whether or not a reviewer is a valuable person to follow. Their credentials as a reviewer targeted to the viewer’s favorite genre are displayed prominently next to their name, and a grid of book covers they have read and enjoyed are displayed adjacent. To maximize the chances that a viewer will recognize a book, we choose the most popular books that are within the viewer’s favorite genres. If a user wishes to glean more information regarding the reviewer’s tastes, an overlay of the text review can be displayed via a tap-overlay.
This display, however, proved too information rich for practical purposes. Other research indicated to us we needed users to have at least five social connections upon joining Goodreads. Based on business needs, I proposed inserting a still more abbreviated component in a list-view format during account-creation onboarding:
I maintained much of the original design, even in a list view format.
Note that the ‘top reviewer’ endorsement is abbreviated in favor of displaying two matching genres. The covers, while smaller, are still legible on even small phones.
The review text overlay was purposefully omitted, so as to minimize friction in completing the onboarding flow. We found that when there were too many browsing options during first time experience new members tended to drop off before completing all steps, and led to lower retention.
Follow buttons have also been eschewed in favor of auto-checked checkboxes, to minimize friction. While I oppose this UI pattern in ‘Add-Contacts’ invite flows, I deemed it appropriate in this context, since it is our recommended and best way of experiencing the platform.
This experience launched as part of an A/B test in November 2018, and since then we have observed increased activities from users during their first session. However, we have not noticed any notable change in retention. So, while the experience has made a user’s first-time experience more engaging, we have not introduced any new mechanisms that re-engage a user after their first session. Follow-up work on this project will involve developing email notifications, triggered when a reviewer posts content we determine to meet a threshold of interest for our members.