Tuesday, 29 June 2010

What is a data warehouse? And do you need one?

The language of business intelligence can be confusing. Cubes, data warehouses, OLAP, and data mining are all terms that are not exactly self-explanatory. One of the most often used terms in business intelligence is the Data Warehouse, which conjures up images of vast spaces filled with digits. As if 5’s and 8’s all had their own bin in a super-efficient warehouse.

So what is a data warehouse, and how is it different from other databases?

Just about every company has at least one transactional database, and most have many. They store accounts data, contacts, stock or project data. Transactional databases are the ones we use to run our businesses:
  • Those that get updated on an hourly, daily or weekly basis
  • The systems (whether we recognise them as databases or not) that we could not do without.
But not every company has a data warehouse. They are often considered the domain of very large companies, even though that is not necessarily true.

A data warehouse holds historical information. It’s where the data goes after it’s been used in a transactional database system.

As an example, a hotel reservation system is used to let customers know whether there is availability for their preferred dates, and to produce an invoice for hotel services used during their stay. A data warehouse for the same hotel might hold this information summarised by day, month and season so as to better understand customer booking behaviour.

Transactional systems hold detailed information such as the alarm call time for the guest, whereas a data warehouse summarises several years’ data to get a more accurate picture of how promotions or seasonality affect bookings.

In addition, data warehouses can bring data together from several different transactional systems to gain new insights into a particular problem. In the hotel example, costing information might be added to find out which customers are most profitable.

So whether or not you need a data warehouse depends on what your business priorities are. Whether, for example, you want to:
  • Better understand customer behaviour
  • Understand which customer segments are most profitable
  • Send more appropriate marketing communications to your customers
There are many, many more uses for data warehouses, but understanding customers and profitability better is a good start if you haven’t started planning your data warehouse.

No comments:

Post a comment