A few years ago, there wasn't a solution. A classic, state-of-the-art retail chain looked something like this:

At the store
A sophisticated POS system collecting sales data down to the line-item level.

At the Head Office
A mainframe, or AS/400, which collected the sales data via the communications network, and then ran large retail applications, notably Sales Audit, Stock Control, and Merchandising. The output from these applications was reports - either huge (and unread) or heavily summarized. It was possible to have online access to the sales data, but this was usually limited to screens of text, displaying a single item. Looking through it for patterns of behavior was very laborious; it was easier to go to a store and talk to the manager.

Technology has moved on since then.

It has now become cost effective to install a Server at Head Office, either as a replacement for the mainframe, or more usually, as an addition. In fact, the low cost of modern servers has made it feasible to put a server in each store, if appropriate. If we put a modern Database Management System on the server, we now have a very efficient and flexible tool for interrogating very large quantities of data, with the objective of discovering useful information. This technique is currently referred-to as Data Mining.

However, if all the server did with the data was generate large reports, like the mainframe, it wouldn't do you a lot of good.
In parallel with the advances in Servers, PCs have become far better at displaying graphical data - images which used to require special workstations costing tens of thousands of dollars are now available on standard PCs. Thus, the PC has become a very effective client workstation.

There's still one more piece to complete the puzzle: the application.