REAL "AVAILABILITY" CAN ONLY truly be measured by the end-user of a system, as ultimately any problem that affects a user's ability to transact--whether the bank's own system problem, or the fault of Internet Service Providers (ISP's), or even the client's own hardware--affects their business.
Generally speaking there is a huge gap between the users' perceptions of "availability" and that which can be measured by IT groups within the company. Measuring availability from strictly inside the IT infrastructure tends to result in best-case scenario measurements, as it igno