Helping people turn memories into written stories.
Vona is an AI storytelling platform that helps people turn memories, their own or their loved ones’, into written stories.
Story creation is both Vona’s core value and its onboarding flow. However, a large percentage of users failed to complete their first story, limiting activation and growth. I redesigned onboarding to reduce early friction, help users start writing faster, and increase story creation.
Project Impact -
Completion rate increased from ~50% to ~80%
Stories created per month increased by ~30%
Results sustained month over month
Why This Problem
Vona’s primary KPI is number of stories created.
We evaluated several growth levers (traffic, retention, content quality), but funnel analysis showed the largest loss occurred during story creation. Improving onboarding was the highest-leverage opportunity to move the core KPI.
The issue became urgent after a sustained ~60% drop-off during story creation, combined with leadership pressure to improve this metric.
What Was Broken
The original onboarding consisted of a five-step story creation flow.
Two steps accounted for most of the drop-off:
Step 1: ~20% of users dropped immediately
Step 2: >50% dropped after partial progress
Interestingly, users who past step 2 were highly likely to complete the process. The problem wasn’t effort, it was hesitation at specific moments.
Key Insights
I ran funnel analysis and interviewed 10 users (new and existing).
Three insights reshaped the problem:
Users were asked to decide too much, too early.
Naming projects, defining characters, and choosing formats before understanding the outcome created anxiety.
Requests for “more clarity” masked uncertainty, not confusion.
Behavior showed users stalled when asked to commit without confidence, not when information was missing.
Starting was the hardest part.
Once users felt momentum, they were willing to continue.
What I chose not to fix
I intentionally kept story setup steps like language and genre selection. Research showed that when AI acted without explicit confirmation, users felt a loss of control. A lightweight confirmation preserved trust before triggering generation.
Design Strategy
Primary goal: Reduce time-to-value
Supporting goal: Reduce cognitive load
To achieve this, I delayed non-essential decisions and removed upfront explanations about story structure and scope. This meant accepting less initial clarity, and less context for AI models, in exchange for faster engagement through writing.
Results
The most significant improvement was completion rate, which had direct affect on number of stories created. This directly addressed early hesitation, and had been the main blocker.
Completion rate
From ~50% to ~80%
Stories created per month
Increased by ~30%
Results were measured month over month and remained stable within a ~5% margin, with strong confidence in the impact.
giladrfg@gmail.com