The Challenge
What Discover Was Facing
Discover offered a personalised discovery and recommendation engine as a B2C subscription product. Trial sign-ups were healthy, but conversion to paid was poor at 8% because the recommendation quality in the first session was low — the algorithm needed enough interaction data to produce meaningful results, but most users hit the paywall before accumulating it. The product's core value was invisible during the trial window.
The Solution
What We Built
We redesigned the onboarding to front-load preference collection through an engaging quiz that seeded the recommendation model before the user saw their first result. Collaborative filtering from anonymised cohort data was used to warm-start new user profiles, ensuring quality recommendations from session one. The trial length was extended to 21 days with a progressive paywall: basic features remained free indefinitely, premium recommendations required a subscription. A reactivation email sequence targeting lapsed trial users recaptured 18% of previously lost leads.
Results