The problem with this, if I understand what you mean correctly, is that an account that expired and or received paid time during the last 30 days may be counted both as active paid and active free (or, possibly, as active free, active paid, active premium paid), which could cause bias of another kind.
(And FTR, your SQL isn't silly. Implementation-wise, if I were to use your scheme, I would probably do a join with the existing account level history table which IIRC exists somewhere, rather than duplicate its content in clustertrack2. But if that table didn't exist, your implementation would make sense.)
no subject
(And FTR, your SQL isn't silly. Implementation-wise, if I were to use your scheme, I would probably do a join with the existing account level history table which IIRC exists somewhere, rather than duplicate its content in clustertrack2. But if that table didn't exist, your implementation would make sense.)