Disciple · Case study 01
There was an existing implementation of messages with very low usage and basic features. How could we increase usage while maintaining the feeling of a safe space?
The scene
Michael Harrower
Head of Product Design · Disciple
A prominent JTBD for Disciple customers was their desire for a safe space for their members. A lot of communities were built around discussing personal, delicate subjects that could make the end user (member) feel too vulnerable to share their experiences via a public post.
On paper, messaging seemed like important functionality for communities that thrived on member privacy and safety, but it wasn't being used. There were a few quirks about the current experience that might have been to blame.
The problem
We needed to discover the biggest barriers causing the low usage in the feature and try to remove them.
Very low usage
Members weren't connecting via messages
94
Feedbacks submitted via customer portal
The work
The project was split into four phases: defining scope and the biggest value levers, researching messaging UX, prototyping with the engineering team, and implementation.
Phase 01
My first step was to use Claude to synthesise all the available data from the customer feedback portal, user interviews, usability tests, and support tickets. These were the main themes of feedback:
1.Removing the “Connect” Requirement
A very common complaint: members must send a connection request before they can message each other. Admins and hosts especially want to bypass this, so they can message any member directly without waiting for acceptance.
2.Message Deletion & Editing
Users want to be able to delete or edit individual messages. Several hosts noted this is a deal-breaker for their communities (particularly for privacy/sensitive use cases), and one even cited it as a reason a deal didn't close.
3.Group Chat Improvements
Requests include: naming/renaming group chats, adding/removing members after a chat has started, replying directly to a specific message within a group thread, and emoji reactions (thumbs up, etc.).
4.Message Search
Users want to search through past messages rather than scrolling through entire histories.
5.Admin/Host Messaging Powers
Hosts want to send mass/broadcast private messages to all members or segments, and admins want to DM members without needing a prior connection.
6.Read Receipts & Timestamps
Requests for read notifications (to see if a message was read) and for message timestamps to update dynamically.
7.Message Management
Requests to mark messages as unread, archive them, flag/label them, and export messaging data — mainly from admins managing high volumes.
8.Push Notifications as SMS
Some users want push notifications delivered as text messages instead, since members miss app notifications.
9.Minor UX Issues
Including: multi-line input on Android not working, inability to paste images from keyboards (e.g. Bitmoji), clickable hyperlinks not working on iOS, and confusing terminology between “connection requests” and “message requests.”
And we knew a lot about our customers already:
Phase 03
After a kick-off meeting and some discussion about complexity of each item with the team, we landed on aiming to solve:
Starting a chat with a member
Deleting a message
Editing a message
Editing a message — time limit
Message privacy & safety
Managing archived chats and blocked members
Group chat management
The outcome
6wks
Time to ship
+153%
Increase in Messages feature usage
−82%
Reduction in related feedback portal and support tickets
In 6 weeks, we addressed a large number of the feedback identified.
What I learned
Claude's ability to synthesise data accelerates research
Being able to synthesise hundreds of tickets and feedback items, and 50+ interviews and usability tests in such a short time is a huge efficiency boost. It's also, when used wisely, bound to reduce memory bias.
Removing barriers to connection doesn't always mean less privacy.
With some thought into how to give the members control over who can message them, we were able to increase connection without reducing the feeling of a safe space. It helps, of course, that Disciple communities are close-knit by nature.
Next case study
Prospects were leaving before they saw the product. We needed to reduce commitment for the user and embrace learnings over perfectionism.
