Subversive · Nov 27How Preply Built the World’s Largest Language Learning Marketplace
Preply defines success as a long-lasting tutor relationship, measured by subscription, and has run hundreds of experiments against that metric. A deeper onboarding questionnaire gave its biggest conversion win, and explaining why a tutor fits a learner was among the best matching wins. The growth team is about 10 people in 10 product squads.
Tips
- Add a deeper onboarding questionnaire to filter for high-intent users▶ 24:51The conversion win was like nothing Simon had seen before, despite his travel background where friction is bad.
- Measure growth experiments against subscriptions, not funnel proxy metrics▶ 27:45Preply has run hundreds of experiments judged this way, and wouldn't claim success on a change that raises registrations but not subscriptions.
- Define success as long-lasting relationships, measured by subscription▶ 16:41Every matching improvement is checked against this subscription success metric via experimentation.
- Explain to learners why a tutor is a good match for them▶ 18:38Simon names it among the best wins in the matching funnel.
- Use reviews from similar learners to build confidence in a match▶ 19:36Simon cites deeper use of user-generated content as one source of great success in improving matching.
- Build matching on three parts: learner data, tutor data and an intelligence layer▶ 12:46
- Optimize the trial lesson itself, not just the funnel around it▶ 30:02Simon calls tackling the trial lesson experience a major unlock in acquisition.
- Cut time to value by surfacing tutor availability and easy scheduling▶ 33:20
- Reduce learner anxiety by showing what a class looks like▶ 36:58
- Work backwards from the real value to define time to value▶ 35:18
- Use AI for subtle personalization rather than chatbots everywhere▶ 44:00
- Raise the hiring bar and expect candidates to use AI▶ 47:10
- Lead so every team member gives the same answer to where and why▶ 55:26
- Expect to be wrong half the time, but know why▶ 58:31
- Start from the most obvious wins in experiments▶ 18:05
- Seek opposing viewpoints to avoid a single way of doing product▶ 52:50
