Voice preview in onboarding vs survey-first
Hypothesis: hearing a persona voice clip before answering the personalize survey raises D1 retention.
Variants. Result. Principle.
LabsWhat we tested.
What we tested.
What won.
Every week we run experiments on onboarding, pricing, lesson structure, and notification cadence. This page is where we publish what we tried, what worked, and what we learned. No spin. Numbers and the principle behind them.
Coming soonFormat preview Format preview
Weekly write-ups post-launch.
Public launch is September 2026. Once we have real users and real data, expect a new entry here every week. Until then, we are shipping the infrastructure and running internal experiments against a private cohort.
Streak-saver framing: gentle vs urgent
Hypothesis: a teacher-framed streak-saver outperforms the industry-default urgent framing on save-rate AND 7-day retention.
Variants. Result. Principle.
Why we publish this
Discipline is the moat.
Most coaching products tell you what they think works. We are going to show you what we measured. If a tactic does not survive contact with the data, we drop it. If a principle holds across experiments, we keep it. The cadence is the product.