N hours of actual use, in isolation, is just the matter of calculating average power draw[W] by runtime[hr] and buying the battery with Wh figures comfortably bigger than that.

e.g. your device consumed 1 Watt on average, you wanted 8 hour runtime, then you need a battery with 8 Watt-hours, or 2,162.162162162162 mAh at 3.7V of capacity, before factoring in buffers of various kinds. But it's also roughly the datasheet nominal capacity of a single 18650 cell.

You don't worry about daily battery life on mainstream mobile devices and you can fall asleep with YouTube playing and it's still playing when you wake up because manufacturers know consumers do that and optimize the phone to make that work. They probably reduce display brightness, cut powers to mics and P cores, ask 3M to make the pouch films 1% thinner so battery could be few more percent bigger inside, fudge battery gauges so you would be nudged correctly to have enough charge before you fall asleep, the list goes as far as your imagination could possibly go.

The fact that same behaviors don't happen on Linux devices, even with something like four of fresh 18650s, means the list ends before it begins. They probably don't do ANY power profiling AT ALL. I'm sure they don't do ANY environmental testing, either.

Would I accept that as a consumer? No. Would I if I was the manufacturer? ...