Business Intelligence
At the heart of every business is data. Whether it is a bank, a shop or a dentist, the things important to that business must be recorded in some way, to track a customer’s balance, stock levels or dentist appointments.
The systems used to store this information need to be pragmatic in nature: ideally they allow employees to enter and edit data in an efficient manner, and give them the information they need to carry out their work now, today. A dentist appointment system, for example, would likely show a calendar of appointments, where a secretary can efficiently book people into slots.
There may be another system, again a pragmatic input system, for the dentist to input results for each appointment.
These systems, by design, do not give any deeper insight into the business data recorded beyond that needed to carry out immediate actions.
But there is great benefit in creating a deeper, analytical view of your data. Instead of day to day concerns, when a longer-term view is taken, trends and patterns can be identified, and actions taken to change those trends to the advantage of the business. This is an extremely valuable tool for any business to have. Like driving a car, the further a driver looks ahead down the road, the more intelligent the driver can be about how they drive, avoiding obstacles and identifying the best lane to be in.
Because of it’s ability to allow intelligent decisions, this type of data analysis is generally called Business Intelligence (BI). It is a huge and thriving industry, due to it’s potential for improving profitability and providing a competitive advantage.
The approach that BI analysis takes is to identify important measures, and to ‘slice and dice’ measures by dimensions. For a dentist, an important measure may be the average number of appointments. This measure could be sliced by the day of week:
| Day of week | Avg. number of appointments |
| Monday | 21 |
| Tuesday | 20 |
| Wednesday | 19 |
| Thursday | 20 |
| Friday | 16 |
| Saturday | 31 |
This is a simple analysis which does not tell the dentist much beyond the fact that more customers are seen on Saturday. But if we dice the measure using a second dimension, age of customer:
| Day of week | Under 18 | 18-60 | Over 60 |
| Monday | 7 | 7 | 7 |
| Tuesday | 7 | 6 | 7 |
| Wednesday | 6 | 6 | 7 |
| Thursday | 6 | 7 | 7 |
| Friday | 5 | 6 | 5 |
| Saturday | 15 | 9 | 7 |
We discover that most of the extra customers seen on Saturdays are under 18s. This may be valuable information to the dental practice, which could guide decisions about the content of their next advertising campaign to target this demographic.
This is a simplistic, contrived example to demonstrate a principle. In practice, the following factors make BI implementation technically challenging:
- Potentially many possible measures and dimensions.
- Potentially gigabytes or terabytes of data.
- Distribution: there may be many disparate data sources.
- Performance: repeatedly running complex multidimensional queries can slow down the data sources, which are needed for business transactions.
A solution to these difficulties, which has been implemented in a number of ways by different BI technology vendors, involves having a consolidated, pre-processed repository of data from all sources, clearly arranged into measures and dimensions ready for analysis. This solves many of the problems facing BI: by pre-processing and consolidating data into one place (often called a data warehouse), we can enable fast and efficient ad-hoc multi-dimensional analysis.
Traditionally, the process of creating a data warehouse is an intensive process, resulting in a warehouse which physically resides somewhere.
However newer technologies are emerging, such as QlikView, and Microsoft’s PowerPivot. These new technologies are changing the way BI can be implemented by taking advantage of the memory and processing power available on today’s desktop PC to erode the concept of a traditional data warehouse residing on a central server, and instead relying on in-memory processing of data on the fly. BI solutions are now easier than ever to implement, which means that you can benefit by having more relevant data upon which to base your decisions.
