Projects

Many group chats involve too much 'irrelevant' chatter which reduces engagement.

My goal was to improve retention and engagement with group chats, which I did by designing a “smart mute” feature which distinguishes message relevance.

🔍 Research
🧠 Ideation
📱 Prototype
🎨 UI Design
🧪 Testing
☁  Personal Project
See the Final Results
🏆

Process & Reasoning

A questionnaire validated the problem and narrowed the scope.

I chose to use a questionnaire because WhatsApp is very widely used affording a sizable response. I got 48 responses - 54% voted "too many notifications" as the biggest frustration with WhatsApp - those individuals rated Whatsapp Group Chats 6.7 out of 10 for satisfaction (on average).

Expand

How do all the other chat apps address this problem?

I wanted to run competitive analysis because there’s a huge number of chat apps available with different features and options. All competitor solutions that I found required effort or an advanced product knowledge. There was no quick and easy fix available to block out the noise. I also interviewed 6 users but didn’t really uncover much additional insight.

Expand

Redefined Problem

Users need a simple way to block out irrelevant messages, because the ratio of relevant-to-irrelevant messages is directly proportional to chat engagement.

Discarded Solution

More group-admin control - but then all participants are relying on someone else to shut everyone up... we need all users to be in control.

Expand
Discarded Solution

Breakout Groups - again everyone relies on the admin to arrange, and too much effort for the benefits. It's easier to just ignore or mute the chat...

Expand
Discarded Solution

Nested Messages - this partially solves the problem on platforms like Facebook except again, you're relying on other people to use the feature so you don't get spammed.

Expand

Chosen Solution

"Smart Mute" would allow notification exceptions for keywords such as: the user's name, dates, times, email addresses, phone numbers, "Everyone", "Anyone", "you all" etc. And to ensure an easy setup, this mode would auto-enable on a chat that that you frequently ignore.

But why did I choose THIS solution?

The research showed that message relevance and user involvement is directly proportional to engagement. An assumption is that these keywords and phrases are the most relevant parts of an overly chatty group.

**This was a big assumption that I would research thoroughly first, in retrospect.

Expand

Two design principles that I wanted to adhere to...

Defining message “relevance” when everyone values their contacts and topics of conversation differently is complicated. In my MVP I defined it based on some English keywords and phraseology (see "Going Forward" to see how I'd level-up this method).

The solution would also need a super simple setup because otherwise people wouldn’t bother using it or setting it up on multiple chats.

Expand
See iterations
📱>📱

User Testing & Iteration

Users overlooked or didn't read / understand the initial notification.

The first design would show an inline message if a loud group had been "smart muted". Most users dismissed it instantly or didn't understand very well. To solve this I replaced the wordy inline prompt with a simple "muted" icon.

Before
After

Most users hesitated when toggling the mode ON or OFF.

Turning Smart Mute ON and OFF in was initially a toggle switch. Users hesitated on this screen revealing a perceived double negative form the toggle switch. Ultimately I moved the Smart Mute feature into the existing mute wizard as a very simple checkbox. In the second round of testing, all users completed the task without hesitation and with a sufficient understanding of the feature.

Before
After

Final Results

The final solution reduces nuisance and irrelevant notifications with an easy setup, whilst still offering full control to the user.

Users needed a way to hear only relevant messages, because message relevance & user involvement are directly proportional to enjoyment. An automatically triggered mute that makes exceptions for keywords and rich data addresses this problem.If this was a real feature I would measure adoption and retention: do users actively enable the feature on chats? Do they keep it enabled?

Going forward...

Future versions might incorporate machine-learning by addressing user behaviour patterns in relation to the content of incoming messages.

...Lessons Learned

In hindsight I would do a LOAD of research on what defines message relevance. I was a young designer when I did this project, please forgive my oversight!