It's a bit risky to extrapolate total battery use based on the battery indicator after 3 minutes, the rate at which it depletes could be not constant, and there are reports of the calibration being off... And especially if your precision stops at the percent (rounded off, so there can be an error of 1~2 percent, and your measure is 3~4%). You haven't measured time with a stopwatch for how long it actually took either, so it's a rule of thumb applied over approximations.
Edit:
to top it off, apparently the battery indication itself is "an educated guess" and not an actual "measure" of how much battery there is left.
Since you can't measure voltage direcly like an alkaline pile, the program guesses based on past consuption how much should be left (until you empty/charge it completely to re-calibrate it). Basically they
have a model of how a battery should behave based on temperature/number of usage, and hopefully the data collected fits with the model.
Didn't do the math, didn't do the science (write down measurements). Not thread worthy. Difficult to have any intelligent conversation. I hope someone comes back with a better test to start anew.