What is end-to-end latency in a smart-glasses app?
End-to-end latency is the time between a defined starting event and the useful result reaching the wearer. For a voice question, that might run from the end of speech to a readable answer. For a notification, it might run from the source event to visible text in the glasses.
Define both ends before timing
Changing the start or finish changes the metric. Starting after audio upload excludes part of the user's wait. Stopping when a server returns excludes delivery and rendering. Neither measurement is inherently useless, but it should be named for the portion of the workflow it actually covers.
We use separate labels for total user-visible delay and internal stage timings. That lets a team identify bottlenecks without presenting a fast server response as proof that the complete glasses experience is equally fast.