Drop one script tag into your app, give someone a task, and get it all back: the screen replay, their face, roughly where they looked, every click and pause — and a plain-English read on where it went wrong.
Time on task, clicks required against the shortest possible path, and where the wasted effort went.
Hesitations, rage clicks, clicks that did nothing, backtracking and mouse thrash — scored into a confusion timeline you can scrub to, then read back against their face.
One participant is an anecdote. Watching U aggregates every attempt at a task so you can tell a real pattern from a bad run.
Normal visitors load about a kilobyte and nothing happens. The recorder only wakes up when a study link is used.
Watching U records the participant's face so you can watch their reaction next to the replay. That's real personal data, so it's handled like it: participants see exactly what will be recorded before anything starts, there's a stop-and-delete button on screen the entire time, recordings are deleted automatically on your retention schedule, and you can drop the video while keeping the analytics. Typed characters are never recorded at all — only how much someone typed and how often they corrected themselves.