Zoom at Scale: Latency Mitigation and UI Patterns for 10,000-Participant Events

Tech · 7 min read

Zoom's approach to large-scale events combines server-side mixing for public streams with seeded low-latency channels for panelists. The teardown breaks down the architecture: multiple transcode layers, edge caching, and selective forwarding for gallery views. This hybrid topology reduces end-user bandwidth while keeping real-time interactivity where it matters.

UX changes focused on moderator tools and participant affordances. Moderators receive predictive queueing hints for Q&A and automated speaker handoff. For attendees, the product separates passive viewing UIs from interactive panels, simplifying expectations and reducing the cognitive load during large sessions.

Operationally, Zoom's event product emphasizes pre-flight checks, staged rehearsals, and clear fallbacks (dial-in and dedicated livestream URLs). For designers building large-call UIs, the takeaway is to treat passive and active participants as distinct personas and optimize latency differently for each.