Banking... TL's Experience

'Megabytes' used to be HUGE just a decade ago but now there are terabytes and petabytes of data for reasonably large organisation.

Amazon famously analyses the hundreds of terabytes that they get from millions of users on both shopping cart transactions and clicking/browsing  data. Using this analytic engine, they are able to piece together products that are related to each other from the millions of products in their database. They are then able to recommend these based on online behaviour of their customers.

Retail banks face similar situation. With online banking and online marketing activities in addition to their traditional data from branches, callcenter, ATM etc, banks have even bigger amount of data on hand. Analyzing these costs money. How can they handle these cost effeciently?

One way for some is to analyze a smaller set of data that is representative of the entire set.

That's smart. Right?