Throughout almost all of FICO's history, the big three credit bureaus stored only very limited data on credit card balances. Every month the newly reported balances would wipe out the old ones.
This is crucial because it means that FICO had no way to tell what it meant when a person had all cards reporting at zero. It might mean:
(A) That the person had been using credit cards frequently for years, but this most recent month was a fluke.
(B) That the person had been using credit cards frequently for years, including the last couple months, but he always paid the cards to zero.
(C) That the person had not used his cards even once for the last three years.
The CRA gave FICO no way to detect which of the three was the case. Persons A and B are regularly using credit, but never missing a payment and keeping their debt very low. Person C is not using credit at all, and hence his absence of lates and his ability to keep debt low is much less meaningful.
Because FICO lacked the ability to distinguish the three scenarios, it had to rely on a proxy. The developers marked a person as using his cards (A or B) if at least one positive balance reported; otherwise if all cards reported zero it marked him as C.
