What the grid coverage number tells you
Each cell lights up once when a drag crosses it, and the readout counts how many of the 216 cells have been touched so far. The number only means something after a deliberate sweep: drag across the whole surface slowly, covering every row, and watch for a cell that stays dark even though your finger passed over it. A dark cell inside a swept area is the signature of a dead zone, a spot where the digitizer stops reporting contact.
That is the same test Google's Android help page describes for a phone screen that is behaving oddly. Touch the top-left corner, drag slowly toward the bottom-right without lifting, and move slowly enough to count to 10 on the way across. If the notification shade releases partway, repeat the drag and note whether it releases in the same place. Google's instruction at that point is to contact the manufacturer, not to change a setting. Samsung's built-in #0# hardware menu asks for the same thing with a box and an X that only color in where your finger is detected.
One caveat about the count itself: a cell stays lit once it has been touched, so the percentage is a record of where you have been rather than a live map of what is working right now. Sweeping twice over the same strip tells you nothing new. To re-test one region, reset and sweep that region again.
How many fingers the screen can track at once
The multi-touch counter watches every pointer the browser reports, keyed by its pointerId, which stays with a single finger for as long as that finger rests on the glass. Press two, three or four fingers down together and the peak reading records the highest number that were active at the same moment.
That peak is compared against navigator.maxTouchPoints, which MDN defines as "the maximum number of simultaneous touch contact points that are supported by the current device". Desktop machines with no touchscreen report 0. Phones commonly report 5, though the exact figure depends on the hardware and some panels report more. A peak below the reported ceiling usually just means fewer fingers were used, so press more at once before drawing a conclusion. A device that reports 10 points but never registers more than 2, however many fingers you press, is a more interesting result.
The number matters because gestures are built on it: a pinch needs 2 contacts and a three-finger swipe needs 3. When a gesture that should work never fires, the peak reading separates two possibilities. Either the panel is failing to report the extra contacts, or the contacts are fine and something else is blocking the gesture.
The event log, and what ghost touch looks like
The log lists the most recent touch events with a timestamp, the event kind, the pointer type and the pointer id. It exists for the opposite problem to a dead zone: a screen registering contact that nobody made. Google's Android troubleshooting page lists a screen that "jumps" among the faults it covers, which is the same family of behavior, and the informal name for it is ghost touch, or phantom touch. It is hard to catch because the stray input usually lands while you are looking at something else. Use the log passively: leave the page open with your hands off the glass and watch the counters. On a healthy screen the "down" count only moves when you touch it, and a count that keeps climbing while nothing is touching the screen is the phenomenon people mean by ghost touch.
Not every cause is hardware. Google's Android troubleshooting page and Apple's iPhone screen guide both open with the same non-hardware steps: remove the case and any screen protector, clean the glass, disconnect accessories, then restart. Samsung's guidance adds that non-genuine screen protectors and cases can interfere with touch near the edges, and that a Touch sensitivity setting under Settings > Display exists for screens with a protector fitted. When the problem survives a restart, safe mode, and the removal of every accessory, Google's page and Apple's both stop offering user-side fixes and send the owner to service.
What this test cannot tell you
The grid measures whether touches arrive, not where the operating system thinks they landed. A screen that registers every touch but acts on the wrong spot, which is a calibration offset, will sweep cleanly and still be wrong. On Windows that offset is corrected through Control Panel > Hardware and Sound > Tablet PC Settings > Calibrate; phones handle it in firmware.
A browser also sees touches only after the operating system has processed them. Touch sampling rate, the rate at which the panel itself scans for contact, is a hardware figure that no web page can read. A panel with a low sampling rate can light every cell here and still feel sluggish in a fast game.
The grid is coarse on purpose. Each cell covers a few millimeters of the surface, which is the right scale for locating a dead region and far too coarse for measuring a single pixel. A dead-pixel test is the tool for that job.
Worked examples
Sweep the whole surface and the readout lands on 216 / 216 (100%): every cell reported a touch, so nothing inside the swept area is failing to register. That is the pass condition.
Sweep the same way and the readout stops at 205 / 216 (94.9%) with a dark run of cells down one side: your finger crossed those cells, so the panel did not report contact there. Sweep that strip again. If the same cells stay dark, it is a repeatable dead zone.
Press three fingers down and the peak reads 3 while the device reports 5. The panel can track more contacts than were used, so nothing is wrong yet. Press five and the peak still reads 3, and the panel is dropping two contacts, which is the pattern behind a multi-finger gesture that refuses to fire.