A privacy-first learning recommendation engine designed around 3 personas — with staged consent built in before any behavioral tracking happens, not bolted on after the fact.
Most recommendation engines earn relevance by tracking everything from day one. I designed EduTribe's the other way round: recommendations get smarter as the learner opts into more signal — never the reverse.
Each stage unlocks a richer recommendation signal — but only after the learner has actively chosen it, with a clear explanation of what that stage adds.
Clear, stage-by-stage opt-in screens explaining exactly what each consent level unlocks and how to step back down.
Course feed, "continue learning" rail, and search results — each adapting to the learner's current consent stage.
A single settings screen where learners can see, change, or revoke their consent stage at any time.
Sensible default consent stages per persona, calibrated to typical comfort with data sharing — always changeable.
Built the staged-consent recommendation feed as a working prototype — each course card surfaces a "Why this?" explanation (e.g. "Top-rated soft-skill pairing for analysts") alongside rating, learner count, and duration, so the system stays legible even as consent stages — and the signal behind each recommendation — deepen.