It takes some time for the chemistry to balance out as others have said. Hysteresis. The voltage is measured right at the cathode and anode. But there is maybe 1 mm at least of chemicals (electrolytes) on top of the cathode and anode. You discharge the chemicals right at the cathode and anode and you measure a voltage there. In reality there is probably a voltage gradient. It takes a bit of time for the chemicals to equilibrate.
Also, the chemical reactions and side reactions are complicated and not as simple as they explained it - that is why batteries have a range of voltages from fully charged to partially charged to discharged completely beyond rechargeable.
You would probably get a better picture if you can measure the voltage across the whole length width and depth of the chemicals (electrolytes, anode, cathode) and then average or integrate the results. The problem with the current voltage meter is that, due to cost constraints which are reasonable, you are only taking a point source.
If you wait 5 to 10 minutes, don't use the battery, give everything a chance to settle, that would probably give you the best measurement. But somebody with some experience could probably make an educated guess, integrating in their mind how hard the battery has been working and basing on their experience, and the guess would probably come pretty close. For example, if I were climbing a hill hard on the throttle for a while, when I get to the top and ride slow, the meter will probably go up one bar. It probably won't go up 2 bars. And if I ride hard like that, I can assume I am not going to get the total range (in distance) from that battery. Things work well enough at least for me.