In six years we've had encoders die, keynotes drop off mid-sentence, venue power trip and platforms fall over. Almost none of those became something the audience remembers, because the response was already written down. This is that playbook.
Second 0: cover the picture
Whatever has failed, the audience must never see black, a frozen frame or a gallery desktop. The first call is always "fade to holding" — the branded "back in a moment" graphic with the music bed. It buys time and it looks intentional. Every show has this graphic loaded on a hot key; if yours doesn't, that's the first thing to fix.
Seconds 0–15: diagnose out loud, once
One person reports on comms in one sentence: what's broken and what they're doing. "Primary encoder dropped, switching to backup." "Sarah's link is gone, wrangler redialling." The caller acknowledges and everyone else stays silent unless they have information. Panic on comms is contagious; a calm voice is too.
Seconds 15–45: execute the rehearsed response
Each likely failure has a pre-agreed response from the risk register: encoder → backup encoder; internet → bonded cellular; remote speaker → holding graphic with their name while the wrangler reconnects, or the presenter takes the next item; platform → standby link emailed to attendees; playback file → skip to the next item and return to it. Because it's been drilled in rehearsal, this is a cue like any other, not an improvisation.
Seconds 45–60: brief the presenter
The presenter is told, via the floor manager or an earpiece, exactly what to say and do: "Sarah's line dropped; say we'll come back to her and introduce the next speaker." Presenters who know the plan handle it gracefully; the audience hears "we've lost Sarah for a moment — let's bring in James and come back to her" and thinks nothing of it. Give every presenter a line for this in the prep session.
When to stop and restart
Rarely. Stopping the stream logs the audience out of players and dumps the recording; restarting means many won't find their way back. Stay live on holding for up to a few minutes while you fix. Beyond that, tell the audience plainly on screen what's happening and when to return, and email them. Honesty keeps more viewers than silence.
Afterwards: the post-mortem
Within 24 hours, while it's fresh: what failed, what the response was, how long the audience was affected, what would have prevented it. Written down, without blame, and turned into a change to the checklist or the kit. The crews that never seem to have problems are the ones who've had them all before and wrote them down.
The kit that makes recovery possible
- Holding graphic on a hot key, with music.
- Backup encoder and backup connection, already running.
- A standby stream destination and the email ready to send.
- Dial-in numbers for every remote speaker.
- Printed run order in case the screens go.
- A rehearsed failure drill in the dress run.
Ask your production company what they'd do if the primary encoder died mid-keynote. If the answer takes more than one sentence, keep asking.