Redundancy is insurance, and like insurance it's cheap compared with the claim. A second encoder and a bonded backup connection add a few hundred pounds to a production; the alternative is your all-hands going black in front of 3,000 employees. This is our working risk register — the things that actually break, in roughly the order they break.
1. Internet (the most likely failure by a distance)
What happens: venue line saturates, guest Wi-Fi throttles, an IT policy blocks the streaming port, or a contractor unplugs the wrong cable. The fix: a dedicated wired circuit for the stream, separate from everything else; a bonded cellular unit as the second path, tested on the day; and encoders that fail over automatically. Connectivity in depth.
2. Remote speaker links
What happens: the keynote's home broadband drops, their laptop updates, their kid starts a download. The fix: a tested backup route (phone dial-in at minimum), their slides run from the gallery so the story can continue, and a holding graphic with their name ready. For critical speakers: pre-record a version as insurance.
3. Encoder or production machine
What happens: software crashes, a machine overheats, a driver decides today's the day. The fix: two encoders running in parallel to the same destination or to a switchable backup ingest; a second production machine mirroring the primary on bigger shows; a hard rule that nothing on the show machines is updated in the 48 hours before the event.
4. Power
What happens: a circuit trips when catering plugs in the urn. The fix: UPS units on every critical machine (encoder, switcher, network switch, router) sized for at least 15 minutes; gallery on a separate circuit from front-of-house where the venue allows it.
5. Audio
What happens: a radio mic battery dies mid-keynote; the stream gets room PA feed and sounds like a swimming pool; a remote speaker's return causes echo. The fix: fresh batteries at every break, a backup handheld mic on stage, a dedicated stream mix with mix-minus returns, and a sound engineer whose only job is listening to the stream. Stream audio explained.
6. Platform and CDN
What happens: the event platform has an outage, or a corporate network blocks the player. The fix: stream to a second destination (an unlisted YouTube or Vimeo link) that can be emailed to attendees in minutes; test the platform on the client's own corporate network in advance.
7. Playback and content
What happens: a VT arrives in the wrong format, a slide deck has fonts missing, the file "was definitely in the folder". The fix: content deadline 48 hours before, everything transcoded to a house format and checked, no content accepted on USB on the morning without a test play.
8. People
What happens: a crew member is ill; the presenter is stuck on a train. The fix: crew depth on shows that matter (a second operator who could take over), a standby presenter briefed, and a run order that anyone can pick up because it's written properly.
Budget rule: redundancy on the internet path and the encoder is never optional on a show with an audience over 100. Everything else scales with what the event is worth to you.