If possible, do the X date report on only your smallest club that doesn’t add up to the right number, and compare everyone on that list to the list you think is correct. A close examination of the differences (sorting both lists by last name helps) should clarify which list is actually correct, if either of them are.
We noticed the same problem. Tried to go back in time, and the number of club members (active and hold) Vin65 reported for a year ago is nothing like the actual number we charted at the time. Would love to have a solution to this.
The solution is to download the whole bloody list back to the beginning of time with every single member, cancelled and on-hold. Then do your own “date math” on the start, cancel, and hold dates.
I wanted to see a plot of club growth from the beginning, this is what I ended up doing.