If you want whatever's connected to the battery to operate right down to the last drop of battery charge, then you must consider its internal resistance when it's nearly discharged. That's when attached circuitry will receive the lowest battery terminal voltage.
Commonly we wish to determine which batteries would be suitable for a given application, where we are more interested in knowing at what charge state the battery's terminal voltage becomes too low. For example, a battery which is still 50% charged, but has too much internal resistance to supply the required current and terminal voltage, might not be considered suitable. Here we are more interested in defining some maximum acceptable internal resistance, and finding the charge state corresponding to that.
Vehicles with internal combustion engines keep their own batteries charged, and the battery is only really expected to provide sufficient engine cranking current while it's holding a significant charge. In this case, you might be interested in the internal resistance of a 50% charged battery, using this figure as "worst case".
Sometimes the type of battery is not debatable (perhaps due to size or weight constraints), and the attached circuitry has to be designed around it, not the other way around. In such cases it's more useful to have a graph of internal resistance at all charge states, and maybe even graphs relating internal resistance to battery age, or charge cycle count.
Some applications might require a battery to only ever be used when near-fully charged. I am reminded of questions from people who use "ignition fuses" to launch home made rockets. My response is always to consider the internal resistance of an 80% charged battery.
So the answer is, as usual, "it depends".