Salt + Light
Rethinking Riffs: Smoother, More Engaging Conversations
Small group members struggled to stay connected between weekly meetings
Most challenges in life happen midweek, but outside of gatherings, connection dropped off. Texts and prayer requests felt shallow — often answered with an emoji or not at all. Leaders’ attempts to spark discussion saw little response. The sense of closeness built in weekly meetings seemed to vanish during the week. We set out to design a feature that could bridge that gap and foster real midweek engagement.
1.0 Typical weekly flow of a small group
Meaningful conversations that can start anywhere, anytime by anyone in a group.
Create questions/prompts for people to respond to
Any user in the group, not just the leader, can create a riff question.
Smooth Recording
All users can respond to questions by simply recording their response.
Engage with Commenting
Users can keep conversations going by responding to one another's answers.
# of Questions created increased by 43% daily
This is specifically when Version 3 was released.
Group members felt deeper sense of connection
Qualitatively, users felt they were growing in relationships, even with people they never met in person.
266% increase in DAU
Version 3 brought more people to the app on a daily basis than the previous months.
How might we help members feel present and connected between gatherings?
Why Audio?
Before prototyping, I studied audio-first products like Clubhouse and Cappuccino and connection-building games like We’re Not Really Strangers.
2.0 Market Research: Clubhouse, Cappuccino, We're Not Really Strangers, 36 Questions to Fall in Love
Across both digital and analog examples, I saw that…
Audio captures emotional nuance (tone, pauses, laughter) that’s hard to pickup in text.
Structured, progressive prompts make people feel more safe to be honest and vulnerable.
With these insights, our team felt confident moving forward with guided audio as our core medium for connection.
V1 — Testing the Concept
Riffs is an interactive audio feature where members can send record responses to questions asked by their group leader. For version 1, we wanted to test if guided prompts included within course material could spark vulnerability and audio connection, so we ran a pilot with a group of 8 members with 1 leader.
Key design choices:
3.0 Record to unlock member recordings
3.1 Personal "listen" view for each member
3.2 Only leaders can post questions
3.3 Version 1 flow of Recording and Listening
3.4 Version 1 flow of Recording and Submitting
Insights from V1 Testing
V2: Designing to Address Insights
Design Changes
4.0 Making the Leader's response available for listening to first to reduce pressure and set expectations
4.1 Letting users know upfront that their recording will be shared with others
4.2 Added emoji reactions to increase member engagement with one another
4.3 Version 2 revised full Record > Submit > Listen Flow
Testing
2 Pilot Groups
Outcome
Concerns about privacy and recording anxiety dropped, but engagement stayed low. Members often recorded once but didn’t return to listen to others. Leaders also struggled to create compelling questions.
V3: Removing Friction & Opening Up Conversations
At this point, I became the sole designer on the team. We dogfooded Riffs ourselves, which revealed deeper issues: difficult navigation, one-sided flows, and friction-heavy recording.
Riffs also shifted from content engagement to a conversation-focused feature to increase member engagement.
This required giving all members the ability to create, edit, and manage riff questions. To support different types of conversations, we also introduced distinct question categories with clear visual cues for easier navigation.
Key Design Changes
5.0 Reduced clicks from 3 to 1, making the flow more smooth and lowering the pressure to form a coherent, “perfect” sharing before pressing the “record” button
5.1 Clear categories, “New” tags, and two-column layout make it easier to find and return to conversations, as well as quickly distinguish between types of questions.
5.2 Added Riff comments as a way to foster more conversation beyond a Q&A format. This allowed for more interaction and relationship-building between everyone in the group.
5.3 Riff question and prompt categories to visualize different topics of conversation and spark new ideas of what could be discussed
5.4 Component anatomy of riff card and stage
Testing
Internal Testing
Outcome of Testing and Release
Testing showed smoother recording, easier re-engagement and finding content. We also saw an uptick from 30 to 110 daily active users for a few weeks from Apr 30-May 20th when the V3 went live. The average number of riff questions created increased from 4.6 to 6.6 questions per day.
Some highlights of user reviews on the feature
The most important lesson I learned from working on this project as well at Salt+Light as a whole is that design must align with product strategy. Incremental UX updates to features can’t fix a weak PMF. If I were to do it again, I would prioritize earlier validation and usability testing to make sure we were building something people really wanted sooner.
I’ve written a short post about my learnings from Salt + Light here: [LINK]
Due to overall low engagement and retention in the app, we sunsetted the product after the Riffs release. Overall, I’m proud of how we transformed a frustrating experience into something more frustrating and engaging.
See more work
Check out the work I did for Butternut, a web app that uses AI to generate a full website for you.
