OPS · Agency PLAYBOOK

The incident response SOP: stream crashes, claim incidents, and bad nights

Updated 2026-10-16 · Sources & methodology noted inline

THE DESIGN PHILOSOPHY: an incident is any event that threatens revenue, compliance standing, or reputation mid-session or after — a crashed stream, a claims mistake said on camera, a hostile brigade, an oversell, a safety question you answered badly. Incidents are not improbable; they are scheduled, the only variable is whether your response is a rehearsed procedure or an improvisation in front of a live audience. This SOP classifies incidents by type, assigns each a first-move protocol and an escalation path, and — the part most operations skip — a post-incident review that converts the event into a checklist change. The goal is not preventing all incidents; it is never having the same incident twice.

CLASS 1 — TECHNICAL (CRASH, FREEZE, AUDIO LOSS): first move is scripted and timed: acknowledge in one sentence if partially live ('we're having a connection issue, back in one minute'), restart on the backup device per the studio SOP's redundancy spec within three minutes, and on return explain once, briefly, then resume the loop where it broke — never restart the session's energy from zero with apologies. If the outage exceeds ten minutes, post to your community channel (the email/SMS list from the marketing compliance sheet exists precisely for this) with the reschedule, and log the incident with duration and suspected cause. Escalation: repeated Class 1s in a week means infrastructure review, not host blame — the post-incident review asks what the system failed to provide, because streams don't crash, setups do.