I just read that in the Thirty Years war about 8 million people died and it got me wondering just how they figured out that number, since I assume no one was keeping a specific individual casualty count.
It depends on what the person making the estimate had available. It's usually done by cross-referencing a number of different sources, including (but by no means limited to): tax revenues, army sizes (estimating these is whole another topis), proportional increase in legal documentation concerning inheritance, percentage of nobles dead (there are pretty good records about these), amount of goods in trade (especially alcohol or food trade is telling), archaeological evidence and careful critical reading of period chronicles.
If you have enough of these, you can come up with a reasonable estimate, and the less you have, the less reliable your figure is.
And the records of how many people lived where are better during 30 Years war than you think - high medieval and earlier have almost no direct data, but the further on you go, the more you have to work with, for a number of reasons.