Hammering out data -governance policies and getting executives to buy-in are two of the thorniest issues when companies extend business intelligence throughout the organisation, according to several users.
Companies want to move away from using BI tools as a way for a handful of users to analyse historical data to, instead, giving frontline workers access to the tools. But building common data definitions — or metadata — has proved a challenge for some organisations.
MC Sankar, vice president of enterprise applications development at TD Ameritrade, says his company is using a BI competency centre to help various business units reach a common definition of how they characterise data.
“The business units have been very satisfied working in silos,” he says. “Without [common metadata], you can be looking at the same data and be making two different decisions.”
This common metadata model will be key as the company moves to marry its BI analysis into the processes of its frontline workers, such as call centre representatives, he says.
Becky Wanta, global chief technology officer at PepsiCo, says the soft drink maker is in its third year of an IT transformation project to tie together six different architectures. This includes the building of an enterprise data warehouse that will function as a “single version of the truth” across PepsiCo’s various divisions and brands, she says.
Key to the success of the data warehouse project, which first went into production in January, was assigning a single owner to the various categories of data and data systems, such as customers, products and shipments, to help build its common data model, she says. This model will be used for data which is pulled from 1,056 sources and 932 applications, according to Wanta.
“This is not a technology play,” she says. “If the business isn’t the owner of all this information you end up with a mapping mess.”
In addition, PepsiCo is using an enterprise service bus and web services to help connect the disparate systems and applications without having to hard-code with point-to-point integrations, Wanta says.
“The enterprise data warehouse becomes the cornerstone, if it is designed right, to be the single version of the truth ... by using your connectivity strategy,” she says.
Healthcare provider McKesson is also working on a data governance structure as part of its BI efforts, says Brian Hickie, vice president of business intelligence at the company. McKesson has also tapped business owners, such as those who work on the sales-to-cash process, as the owners and stewards of that data, he says.
The company’s effort to fuse its processes with BI data has helped users better understand where they fit into those processes, Hickie says.
“The decisions are now being based ... on a more comprehensive understanding of that process,” he says.
An impromptu poll of users at the recent Computerworld US BI Perspectives conference showed that 40% of those surveyed believe that effective management of change control, including the consistency of data and common data definitions, is one of the most typical challenges to the “democratisation of BI”. Another poll revealed that 43% of users think top management’s lack of understanding of the value of BI is the biggest cultural barrier to BI adoption, while 25% say departmental silos in their organisations and a lack of BI competency centre are the biggest impediments.
Companies have taken various paths to gain management acceptance of their BI projects. At 1-800-Flowers.com, for example, a handful of initial quick successes were key to garnering executive support, says Enzo Macali, senior vice president and CIO.
In a demonstration at the conference, Micali tapped five metrics, such as gross margin and profitability, which executives previously didn’t have access to before the BI project.
“Pick a handful of metrics that are meaningful, [and] they will ask more questions,” he says.
Gregory Corrigan, vice president at commercial fleet management services provider PHH Arval, says the company is using BI analysis to show its clients how it is meeting its service-level agreements. In the past six months, the company has begun using BI data to predict when a vehicle is likely to break down and how much it may cost for repairs, so customers can take steps to avoid that cost, Corrigan says.