Kajabi Communities
Rebuilding a community platform for new creators
My self-started exploration became the company's Q4 launch
I was the lead designer on Communities, the only designer on this product. I owned it from research through delivery.
Strategy
Research
Design Systems
Delivery
Three ways creators stalled
We had almost no behavioral analytics, so I ran five unmoderated tests and sat in a weekly creator loop while we redesigned.
“I’m in. What now?”
No next step. Nothing to do, nothing explaining why to do it.
“Doesn’t reflect my brand”
No color, no image, no way to make it theirs.
“I can’t find Live”
Channels, chat, and feed below the fold. Secondary features held the top bar.
Three decisions, and what lost.
A hard launch date meant the rewrite covered two surfaces. Navigation and feed were rebuilt, everything else was reskinned.
Option A
Moved features left, kept the existing groupings
Result
Least disruption, kept the model that was already failing
Option B
Embedded features inside each channel
Result
Discoverable in context, invisible to a creator with no channels
Final Design
One rail, grouped by creator intent, ordered by first-session frequency
Grouped by creator intent
Live room persistent
Active state always visible
Fewer items, easier to find
Feed and chat were scoped as two workstreams.
Creators were abandoning feed to get warmth and giving up feed's features to do it, so I built one component instead of two systems.
Feed
Chat
I designed the unconstrained version first, months before it was scoped. Engineering costed it against patching the legacy architecture and both carried a significant cost. That made structural work worth considering, although we still couldn't take on the entire vision.
Before
Full of features, no place to start
Multiple entry points, no clear start.
Actions buried
Core actions hidden from first-time creators
Empty states silent
No guidance when content is missing
Generic branding
Every community looked like every other community
After
One start, one rail, a page creators can brand
One clear starting point, reinforced across navigation and empty states.
Unified navigation
One mental model → "Structure before surface"
Clear CTAs
Guided empty states
→ "Teach before content"
Brand at surface
Ownership from first session
→ "Design for 0→1"
Measured impact
- 3 pp
Target
- 4.2 pp
Actual
30-day churn
75 %
Target
82 %
Actual
First-session post
Creators now call Communities
“the easiest way to keep subscribers engaged.”
Creator feedback
Thank you.
Kajabi Communities
Rebuilding a community platform for new creators
My self-started exploration became the company's Q4 launch
I was the lead designer on Communities, the only designer on this product. I owned it from research through delivery.
Strategy
Research
Design Systems
Delivery
Three ways creators stalled
We had almost no behavioral analytics, so I ran five unmoderated tests and sat in a weekly creator loop while we redesigned.
“I’m in. What now?”
No next step. Nothing to do, nothing explaining why to do it.
“Doesn’t reflect my brand”
No color, no image, no way to make it theirs.
“I can’t find Live”
Channels, chat, and feed below the fold. Secondary features held the top bar.
Three decisions, and what lost.
A hard launch date meant the rewrite covered two surfaces. Navigation and feed were rebuilt, everything else was reskinned.
Option A
Moved features left, kept the existing groupings
Result
Least disruption, kept the model that was already failing
Option B
Embedded features inside each channel
Result
Discoverable in context, invisible to a creator with no channels
Final Design
One rail, grouped by creator intent, ordered by first-session frequency
Fewer items, easier to find
Live room persistent
Active state always visible
Grouped by creator intent
Feed and chat were scoped as two workstreams.
Creators were abandoning feed to get warmth and giving up feed's features to do it, so I built one component instead of two systems.
Carries depthand structure
Feed
Chat
Feels like a conversation
I designed the unconstrained version first, months before it was scoped. Engineering costed it against patching the legacy architecture and both carried a significant cost. That made structural work worth considering, although we still couldn't take on the entire vision.
Before
Full of features, no place to start
Multiple entry points, no clear start.
Actions buried
Core actions hidden from first-time creators
Empty states silent
No guidance when content was missing
Generic branding
Every community looked like every other community
After
One start, one rail, a page creators can brand
One clear starting point, reinforced across navigation and empty states.
Unified navigation
One mental model
→ "Structure before surface"
Clear CTAs
Guided empty states
→ "Teach before content"
Brand at surface
Ownership from first session
→ "Design for 0→1"
Measured impact
- 3 pp
Target
- 4.2 pp
Actual
30-day churn
75 %
Target
82 %
Actual
First-session post
Creators now call Communities
“the easiest way to keep subscribers engaged.”
, Creator feedback
Thank you.