The team had been A/B testing onboarding for two years. We'd lifted what was liftable through micro changes — copy tweaks, button positions, ordering — and retention had plateaued. Marketing was driving traffic the onboarding couldn't capitalize on, because the flow had no way to reflect the user's actual reason for showing up.

Hypothesis: apply the Hooked framework (Invest → Reward) and personalize value props to the user's stated context. Break the plateau on completion, and create a hand-off point where marketing traffic translates into product fit.

BetterSleep had two distinct user behaviors: sleep-sound listeners who opened the app, hit play, and slept; and sleep trackers who used the recording feature. The features worked in silos. Roughly 80% of users opened the app and immediately hit play — limiting their exposure to anything else we'd built.

The old Home was exploration-led: a grid of content, a player, no narrative. It put the onus on the user to find their own path to value. Meanwhile, anything we promised during onboarding was easy to lose track of once the user landed.

BetterSleep was a cross-functional design lead role. I was the design owner for Lifecycle and Growth, partnered with PMs on hypothesis framing, engineering on feasibility and ACs, and stakeholders on roadmap. Two principles stayed constant:

• Define the terminology before you design. "Engagement," "friction," and "retention" all sound concrete until two people on the same call mean different things by them. The first move on a discovery pass was always to convert abstract words into explicit problem statements.
• Stick to the playbook when you can: Goals → Data → Problem statements → Hypotheses. When the team agreed on that flow, design reviews were short and shipped work didn't get re-debated. When we skipped a step, we paid for it downstream.