Design an audio product for Meta that brings people together

Design a new audio product or feature for Meta that fosters communal listening and connects listeners with creators.

Context

Meta has huge social graphs across Facebook, Instagram and WhatsApp but no strong audio product. Spotify and YouTube dominate listening. Live audio rooms had a short boom and then faded.

Clarifying questions to ask first

Music, podcasts, or live talk?
Which app is the home for this?
Is the goal time spent, connection, or creator monetisation?

How to approach it

Clarify the business goal and tie the product to the company's mission before touching solutions
Segment users by behaviour or need (not just age), cover the whole ecosystem, and pick one segment with explicit criteria
Walk the end-to-end journey and separate real pain points from feature requests
Brainstorm a range of solutions (quick fix, data/ML, moonshot) and prioritise with impact vs effort
Define a North Star, funnel metrics and guardrails, then call out risks, edge cases and mitigations

Framework: Goal & mission -> Ecosystem & segment choice -> Journey & pain points -> Solutions & prioritisation -> Metrics & guardrails -> Risks & edge cases

Sample answer

Listeners (friends who listen together, fans of a creator) and creators (musicians, podcasters, hosts)
Meta's edge is who you listen with, not the catalogue
Live audio faded because discovery was poor and good rooms were rare

Journey: discover > join > listen together > react > follow creator > return
Solutions: shared listening sessions in group chats, creator listening parties with Q&A, audio replies in Stories
Metrics: North Star = weekly co-listening sessions with 2+ friends; guardrails = session quality, creator retention, reports

Shared listening inside group chats and creator listening parties, built on the social graph rather than a catalogue. Measure co-listening sessions, guarded by creator retention and quality signals.

What interviewers look for

Why Meta wins (social graph advantage)
Listener vs creator ecosystem
Learning from failed live audio
Engagement without addiction