Nine lessons on the most under-documented part of any video product: lip-sync. We cover the PTS/DTS clock model, how audio sample-rate conversion produces silent drift over hours of playback, the encoder pre-roll problem, the 40-millisecond human perception threshold for A/V offset, and the production workarounds for live (LL-HLS, WebRTC), VOD (HLS/DASH), and conferencing pipelines. Includes a diagnostic flowchart for the lip-sync bugs that QA finds but engineers can't reproduce.