IF you are loading for precision shooting…
I've bought thousands of new cases over time. The thing I've found is that even with military brass you get some lot-to-lot differences in weight and neck wall runout, and that can happen even when the heads all have the same year stamped on them. There might have been 20 lots run that year. No way to know. So buying them all at once generally gets them from the same lot. That, of course is usually still not highest purity. They typically are the mixed output of several sets of tooling. Even Lapua does that. So, if they are for match loads or other maximum precision testing loads, I lay out a piece of graph paper and mark it in tenths of a grain across the identified weight range and weight the cases to line them up. This results in a sort of histogram that tends to reveal what tools the cases came from by giving you rough bell curves for each tooling set. So the point in weighing is not to match case capacity, but tooling source. I frequently find one set of tools provides less neck wall runout than another and so on. This affects accuracy in some chambers more than others.
Below is a drawing of what a case line-up histogram I made with some Winchester 308 cases looks like. The drawing after it is and Excel histogram made from weights of a smaller sample of some Lapua .308 cases. That saves me making the histogram of cases, but since I am weighing anyway, the cases on graph paper makes more sense as it saves me the Excel entry work.
Now the big question: does this make a clear difference on paper? Occasionally yes, but most often not that I can tell. But me being able to tell isn't my criteria. I know that increased uniformity, while it often makes no obvious improvement, nonetheless cannot hurt, while the reverse is not true. I once found a single Winchester case in a bulk purchase that had 0.004" difference in neck wall thickness across its diameter. A fluke, yes, but in the lot just the same. You could see it clearly by naked eye. I shot that case out in a 10-FP, loading at the range and rotating it 90° at each chambering for eight rounds, then putting it always at 12:00 for 7 more rounds (its neck split on the last round or it would have been 8; I didn't have annealing equipment with me). The former group was just over 1.0 moa, while the latter was just under 0.6 moa. So in that gun's chamber, it would have opened up groups if allowed to position itself randomly, as when feeding from a magazine.
I also know that when I am shooting groups too big for case sorting to make a clear difference, such as from standing offhand, that even though you can't demonstrate an improvement by measuring groups shot that way with uniform cases as compared to groups shot that way without to any serious degree of confidence, that, nonetheless, perhaps only once in 100 or 200 shots or so, that small difference will cause a bullet to scratch a scoring ring that would have just missed scratching it by a tiny fraction without it. So I pick up the very occasional added point. It won't happen often, but why throw away a potential tie breaker? So I continue to sort for matches.