Self-reliance and an inclination for risk taking are two attributes that tend to be associated with the American...
character, but that doesn't mean they don't translate to other cultures.
In the U.K., British Telecom embodies both characteristics when it comes to dealing with customer data. The telecommunications giant's messaging division, with the help of IT consultancy Unisys, built its own enterprise data warehouse from the ground up and decided to take a chance and build open source business intelligence (BI) software on top to create reports and for ad hoc queries.
Every day, BT collects data on the millions of voice, text and video messages its customers send and receive, according to Andre Gayle, messaging common capability service designer for BT. The company wanted access to that data to resolve customer complaints and to identify larger customer use patterns so that it could better tailor its product offerings.
Before the development of the warehouse (which BT calls its Statistical Data Warehouse), messaging data collected by Unisys was basically inaccessible save for weekly, monthly and, only occasionally, daily static reports sent to the appropriate workers by email, said Andrew Hutchinson, BT messaging specialist.
That didn't allow BT the flexibility to quickly respond to customer complaints, Gayle and Hutchinson said. If, for example, a customer complained of losing an important voicemail, BT would have to ask Unisys for the relevant data, unable to rely on the already out-of-date static reports.
"You couldn't wait till the end of the month to hear about something that happened yesterday," Hutchinson said.
Unfortunately, however, Unisys would often return the wrong data, requiring BT to ask for it again. This created a frustrating pattern of back-and-forth communications that resulted in longer resolution times for customers.
So, 18 months ago, BT and Unisys decided to build a data warehouse that would collect all the messaging data the telecommunications company received on a daily basis and find a BI tool that would allow BT's non-BI analysts to access and manipulate the data itself, without having to go through Unisys as a middleman.
With extensive effort, including many custom-built ETL scripts, Unisys built the Statistical Data Warehouse in such a way that it collected all the customer messaging data BT would need and left open the possibility of adding more data feeds as new needs arose. But BT still needed to get at the data.
Hutchinson said BT evaluated a number of commercial BI vendors but found them all to be too rigid in their approach. To support a homegrown data warehouse, he said, BT needed a BI tool that was easily customizable. Not surprisingly, then, the company turned to open source technology.
"With open source, you can customize it as you see fit," Gayle said. Ultimately, BT chose Jaspersoft's open source BI platform to sit on top of the statistical data warehouse.
Using open source software meant that developers at BT and Unisys could tweak the code to meet the needs of users and the data warehouse. For example, BT has strict guidelines as to who can authorize what data, Hutchinson said, and Jaspersoft's open code allowed Unisys to create the mechanisms needed for compliance with those rules.
Commercial BI tools and applications can be customized too, but it is generally a longer and more cumbersome process as it involves making changes to proprietary code, and the vendor itself often insists on doing much of the work.
In addition to ease of customization, BT found that Jaspersoft left a smaller IT footprint than its commercial competitors, meaning that less maintenance was needed, and changes could be made to the reporting and query capabilities more easily in the future.
And, of course -- as with just about any decision to use open source software -- cost was a factor. Though BT would not reveal what it is paying Jaspersoft for the commercial version of its BI software, analysts agree that open source BI costs significantly less than commercial software. There are generally no licensing fees, just support costs.
Lower costs comes with trade-offs, however. With open source technology, companies usually need significant internal expertise to customize the software. Such was the case at BT, with the help of Unisys, but "a small company without expertise may have an issue," Hutchinson said.
Now that the system is up and running, weekly and monthly static reports have been replaced with daily and on-demand reports, as well as the ability for the 50 or so BT workers who use the BI software to perform ad hoc queries to help resolve specific customer complaints.
In addition, BT is able to use the BI applications to monitor network traffic so employees can resolve bottlenecks as they occur and predict where future bottlenecks may crop up, Hutchinson said.
"It gives us the opportunity to look into customer behavior," he said -- a capability that was previously just out of reach.