Latency

Latency is measured one way or as a round trip. Some of it is distance and physics; the rest comes from congestion, queueing in routers and firewalls, satellite or cellular hops, Wi-Fi retries, and the processing inside phones and conferencing systems. The ITU-T recommendation G.114 is the long-standing reference for how much one-way delay a conversation tolerates before it becomes awkward.

For a growing business latency decides whether remote and in-room participants can interrupt each other naturally, and whether an AI voice agent feels responsive or stilted. It also compounds: a slow internet uplink, a busy firewall and a cloud service hosted far away each add a little, and callers feel the total.

Ask your connectivity provider for the latency to the services you use most, not just the speed. Ask your network partner where delay is added on your own equipment and whether QoS is in place. Ask your phone or AI vendor where their service is hosted and how quickly their agent replies after a caller stops speaking.