Your point appears to be (with the latter examples) "well, one says he killed 300, the other says 800 - that means that he definitely killed a shitload, and the exact number is unimportant". I can see where you are going with it - and its a persuasive point - but are you really saying that two sources who disagree quite substantially about the specifics of an event make that event more likely to have occurred?
If we assume there is no difference at all, then things become clear.
In your given example, one counted just men. Another source might
count families. A third source might include slaves. A fourth might
include critically injured who died soon afterwards. A fifth may include
those of that clan. A sixth include visiting relatives. Glad to help.