Aura AI Onboarding
Designing an intelligent onboarding experience that adapted to user behaviour in real time — increasing activation by 47%.
A One-Size-Fits-None Problem
New users were being dropped into a generic 7-step onboarding that assumed the same goals, skill level, and context for everyone. The result: only 31% of users completed onboarding, and those who did rarely returned within 7 days.
Who Are We Designing For?
We conducted 24 in-depth interviews, three rounds of diary studies, and a large-scale segmentation analysis. We identified 5 distinct user archetypes — each with radically different goals, contexts, and mental models around the product.
"I just wanted to get started. The setup felt like it was built for someone else entirely."
Onboarding That Listens
Rather than a linear flow, we designed an adaptive system that inferred user intent from early signals — the questions they asked, the features they explored, and the goals they set. Three distinct onboarding paths branched dynamically based on these signals, each tailored to a different archetype.
Activation Unlocked
After a 6-week phased rollout, the adaptive onboarding outperformed the control across every measured metric. Day-7 retention improved most dramatically — users who completed the new flow were nearly twice as likely to return.
The signal-detection model developed for Aura has since been applied to three other product surfaces, forming the basis of a personalisation strategy across the whole platform.