Time how long the device lasts under a load you choose. Every reading is written down as it arrives, so the result survives the device switching itself off.
This page times what the battery does and records every reading. It cannot tell you the health of the cell — nothing in a browser can.
Pick a load, start the run, and the page keeps the clock and the charge level side by side. What comes out is how long this device lasted under that load — the one battery figure a repair bench can act on.
Every reading is stored the moment it arrives, so a run that ends in shutdown is waiting when the device comes back, finishing at the last reading before the lights went out. That is also the honest way to measure a fresh battery, whose reported percentage is worth less than the elapsed time.
No capacity in mAh, no cycle count, no voltage, no temperature, no health percentage — none of it is exposed to a web page on any platform. Chrome and Edge report the percentage and whether the cable is in, and nothing else.
"How long does it last" needs "doing what". Idle, light and heavy give three different answers, and heavy is the same load the CPU test runs. Two devices only compare under the same load, the same brightness and the same starting level.
Safari and Firefox removed the Battery Status API over fingerprinting, so on an iPhone the page cannot read the level — but it still holds the clock. Six readings typed in across an evening draw the same curve as six hundred.
The level arrives in whole percentage points, so at a light load one reading can be half an hour from the next. The projection follows the last fifteen minutes for that reason.
Charge to full, run it down under a known load, and compare with what the device should manage. A new cell that dies in half the expected time is a bad cell, and the run is the evidence.
Some batteries hold a respectable percentage and then drop dead at 30%. Run to shutdown and the chart shows the cliff, and the level the device gave up at.
Time a charge from a fixed level with the same charger and cable. A curve that stalls, or takes twice as long as its twin, points at the port, the cable or the charging circuit rather than the cell.
Half an hour under a heavy load, watching the percentage fall, tells you more than any battery screen the seller can show you.
Idle drain and drain under load are different faults. A device that lasts all day doing nothing and empties in forty minutes under load has a different problem from one that leaks overnight.
Export as CSV or JSON straight into the job sheet: start level, end level, elapsed time and every reading in between.