A ping to 8.8.8.8 will not tell you why a WhatsApp call from a clinic reception desk in Cheras keeps dropping the patient’s last sentence. The call is not going to that address, and WhatsApp will repair some holes that a ping still counts as failure.
We measure during the app in use. For WhatsApp voice we place and receive calls at the hours the desk is busy. For Teams we join the standing meeting, muted, and watch the same minute the sales lead always loses. For Zoom we ask for a short test room with the same gallery layout the school or clinic already uses.
The numbers we keep are simple on purpose: how many packets never arrived, how long the delay sat above what a conversation can bear, and whether the app recovered by stretching the last syllable or by reconnecting. Recovery is not the same as a clean path.
Staff sometimes want a single percentage they can forward to a boss. A single percentage hides the shape. Ten seconds of heavy loss in a forty-minute meeting can ruin the one agenda item that needed a decision, while a sprinkling of tiny holes across the hour is merely tiring.
Bring your own recordings if you have them. Matching a loss spike to the second someone said “confirm the quantity” is more persuasive in a landlord meeting than a chart with no voices on it.
If you cannot have us in the room, a careful colleague can still help: one person speaks a numbered list, the other writes gaps, and you repeat the list on a wired desk and on the usual Wi-Fi. The difference between those two lists is already a finding.