Bad data destroys BI projects
Up to 80% of large business intelligence (BI) projects that fail do so because of bad data, said Jane Thomson, executive director of EOH, during yesterday's ITWeb BI Summit, in Bryanston.
“Business is not listening to their IT departments when IT asks for help. In addition, most enterprise organisations have complex systems and don't have a single unified data structure design or integrated architecture.”
Thomson stated that more than 75% of enterprises do not have effective master data management strategies.
“People need to make hard decisions about data, and now need to call the shots in terms of how data is managed. New technologies make it easier to deal with large volumes of data. It's imperative that we have linkages that allow us to find the relationships of data coming from different sources,” adds Thomson.
One of the biggest BI challenges is around the issue of data ownership. Thomson indicated that organisations with complex data systems must first define a data and BI strategy, have a recognised portfolio, and secure management's support.
Thomson said it makes sense for BI and information management to work together to deliver results and to fix the data one logical chunk at a time.
“Expensive applications alone do not deliver value. Today we have tools to sort out data issues. Messed up old data plus new technology equals expensive old messed up data,” Thomson noted.
She added that BI challenges are both a data and a people problem. This is often evident in businesses that grow too rapidly without putting the data management foundations in place. “It's all about building a proof case. Find a business champion rather than a technologist; someone who understands systems at a business level.”