Trustworthy user profiles @OfferUp
I launched experiments to iteratively improve UI on the user profile page on the OfferUp app. The goal was to help shoppers evaluate the trustworthiness of sellers more confidently and to help sellers better represent their reputation and personalities.
Time: 2 months
Platform: App (iOS & Android)
Role: Sole designer responsible for user testing, interaction & visual design, and design strategy
Release status: First experiments launched
Metrics: Engagement, retention, visit duration
UI debt makes it difficult for shoppers to determine whether a seller is trustworthy
Scam is the #1 detractor reason for both new and experienced OfferUp users. Besides implementing technical barriers against bad activities, I noted the following on the user profile page during a UI audit:
The layout was cluttered, not allowing the natural eye movement pattern to scan through the page quickly.
The use of color and type sizes and weights was excessive, fighting for user attention and diminishing visibility of important information.
The use of affordances were inconsistent, leaving users confused about what was tappable.
The visual hierarchy didn’t prioritize information based on its importance in determining if a user was worth engaging with.
My team wanted to start iterate on the UI on this page as a low-cost way to improve design hygiene and consistency, collect behavioral metrics to inform bigger feature changes, and set up a/b testing infrastructure.
Profile page of an experienced (left) vs. new (right) seller
Goal
Help buyers evaluate whether a seller is trustworthy easily and confidently, therefore driving engagement and retention.
Seller reputation comes from good buyer experience
Past survey research showed that ratings, reviews, selling history and verifications are most common ways users evaluate trust.
Q: What information is the most helpful for evaluating the trustworthiness of another OU user?Other recommerce and P2P marketplace apps also consistently highlighted these trust signals. In additio, competitors built a sense of trust with free text reviews that provided specifics on past buyer/seller interaction and official endorsements such as badges and ‘star seller’ designations. The most effective visual design tend to leverage white space generously, use size, color and repetition to make emphasis, and center the content around accomplishments of the seller.
Defining a design strategy
I wanted to approach the problem with a few design principles in mind. The content and visual hierarchy should keep user attention on the information that matters the most and facilitate trust between buyers & sellers. The visual design should be minimalistic, use more white space, use as few type sizes as possible, and reserve main brand color to drive attention to the most important trust signals.
In the first phase of my design exploration, I stripped the UI down to grayscale colors and reworked each element to resolve existing affordance and spacing issues.
New and experienced shoppers found it easy to determine seller trustworthiness
With concepts from this phase, I conducted unmoderated user tests to learn how real OfferUp buyers scan, interpret and use information on the profile page in their buying and selling experience.
Research Questions
What trust signals do new and experienced OU users use to evaluate seller trustworthiness?
How easy is it to determine seller trustworthiness with the profile page?
How confident are users about their judgment about seller trustworthiness?
Through usertesting.com, I recruited 4 new and 4 experienced OfferUp users to assume they are looking for a PS5 from three different sellers — one very new, one who have bought and sold a few, and a local gaming shop.
Findings
7 out of 8 participants rated it “very easy” or “easy” to determine whether a seller is trustworthy.
A positive selling history was the most valuable for earning trust for a seller. This meant having a high rating, at least a few items sold, compliments and a join date greater than a month ago.
When there’s little to no selling history to reference, participants turned to other signals such as verifications, profile picture, number of items bought, and other items listed for sale; the more the better.
Flaky behavior was assumed to be part of online secondhand shopping and some participants felt comfortable with engaging even if they were not very confident with the seller’s reputation.
Designing a north star
To plan for the dev work, I worked with my engineer to understand which UI elements were unique to the page and which would involve a global component change and therefore increased QA scope, dependencies and risk.
Knowing that my team was not yet very familiar with running front-end experiments, my PM and I planned the first two experiments to be very small: switching the order of ‘verifications’ and ‘compliments’ section, and redesigning visual hierarchy in the top section and adding an ‘About’ section’. This was so that we could further deemphasize the ‘Follower’ information and prep for future release of custom bio and banner features, based on user test recommendations.
As dev work for the experiments were underway, I began studying how the layout could evolve and accommodate additional features.
When laying out the roadmap for the next year, my PM and I considered a balance between driving engagement and retention directly, adding a sense of neighborliness to the profile, and effort and team’s capacity. For example, even though reviews were highly valuable references for seller reputation, it would require a significant amount of moderation in and beyond the app. We could test and keep developing our moderation capacity through the less complex bio feature and open up reviews to select segments of users.
In my exploration, I focused on making the profile feel more personal and contextual. A significant proportion of sellers and selling activity on OfferUp came from professional sellers whose large posting volume had often been confused as spam/scam. The profile could be an opportunity to visually differentiate these sellers and present official recognition and badges. From a business standpoint, we could consider monetizing with upgraded layouts and customizations with banners and decorations.
Impact & Next steps
The first two experiments were in production waiting for experiment results to reach significance. We expected that shoppers in the experiment group would have similar or better results on engagement.
Metrics included:
% of messages/offers with a response from seller (main engagement metric)
Drop-off after viewing the profile page
Baseline ad metrics.
Once we confirm that the test is no harm, we would be more confident in rolling out more UI improvements and building out POCs for additional features. In the next phase, I will create responsive designs for larger screen sizes to bring parity to the web experience. As OfferUp launched the Jobs feature, profiles would have to represent beyond shoppers and sellers. I will continue to collaborate with other squads on a strategy for reusable profile templates that could fit various types of OfferUp listers.