A metronome tells you where the beat is. Pocket listens through your microphone and tells you where you were — ahead or behind, in milliseconds, and on which beats of the bar. Nothing leaves your browser.
Between the browser scheduling a click and your microphone recording it, anywhere from 5 to 400 ms goes by — soundcard buffers, Bluetooth, the speaker, the air. That delay lands directly on the one number you came for, so Pocket measures it: six sharp clicks with you silent, found again in the recording. Because the reference is the click as your microphone heard it, the result is your timing against the click as you heard it, which is the only fair comparison. On synthetic rooms this recovers the true offset to within about 3 ms.
You need speakers, not headphones — Pocket has to hear its own click. With headphones it will say so, and everything except the ahead/behind number still works, because spread, drift and the per-beat map are all differences, and a constant delay cancels out of a difference.