The test measures how quickly you notice and tap a counter that appears at random intervals. Sleep loss slows the average, and above all makes the slow tail slower: more lapses (taps slower than 355 ms) and more false starts. Response speed is 1 divided by reaction time in seconds, so higher is better and it is less distorted by the occasional very slow tap than mean reaction time is.

Compare yourself against yourself. Same phone, same posture, same time relative to waking, before caffeine. A 3-minute test underestimates real-world impairment because the vigilance decrement grows with time on task, so treat the line as a trend, not a fitness check. Absolute numbers include your phone's touch latency and will not match published lab values.

The performance score is Basner and Dinges' PVT-B summary: the percentage of responses that were neither lapses nor false starts. Their 2011 comparison of PVT outcome metrics found lapses plus false starts, and mean response speed, to be the most sensitive to sleep loss, Response speed is the line to follow: it has no ceiling, so it still separates rested from very rested mornings, where the score sits pinned near 100. False starts have no reaction time and so do not enter speed; they show up in the score.

Charts are positioned by time, so missed days show as gaps. The trend line is a trailing 7-day mean and breaks where more than two weeks are missing. Pausing a test discards the stimulus on screen and stops the clock; the number of pauses is stored with the session. Every session can be edited or deleted later from the Sessions table.

Protocol after Basner, Mollicone & Dinges (2011), PVT-B: 3 min, 1–4 s inter-stimulus interval, lapse ≥ 355 ms, false start = tap before the stimulus or within 100 ms of it.