Big Data Has A Trust Problem

5 Big Wishes For Big Data Deployments

What's stopping your company from getting the most value from its big-data project? A lack of trust might be to blame, according to a new survey from IBM.

One of the report's most interesting findings is that a "trust gap" exists within organizations. According to Kathy Reece, a business analytics leader at IBM Global Business Services, and one of the authors of the study, less than 47 percent of leaders surveyed report a "strong level of trust" between IT and business departments. And when it comes to the C-Suite, the trust gap is even worse: Just 40 percent of respondents feel a strong sense of trust between executives at their firm.

"We've been studying big data and analytics for the past five years now. And this year we surveyed over 900 executives from 70 countries. We wanted to figure out what was triggering analytics adoptions," Reece told InformationWeek in a phone interview.

[ What is in IBM's big data future? Read IBM Answers Big Data Competition. ]

The IBM survey suggests that although big-data projects can deliver big payoffs for organizations, age-old human conflicts often limit their full potential, said Reece. And a lack of interdepartmental trust is highly detrimental to an organization's analytics strategy.

"It also comes down to people trusting each other to share their own customer data from the different lines of business," said Reece. "So if you have information about a customer, and you want to hold it tight to your chest and not share it with the other lines of business, it's really hard for them to get a 360-degree view of the customer."

The IBM study found several examples of how the trust gap hampers business operations. At one "larger bank," for instance, departments would ask IT to complete data-related tasks. "But in the background they would have their own Access database, and they'd be doing the same work the way they wanted to have it done," Reece said.

Companies that fall into this tribal trap are "actually doubling the cost of their analytics workload, as well as duplicating efforts," she added. "So in cost and time, it's impacting them because of the lack of trust."

The key to bridging the trust gap might lie in integrating departments to improve the lines of communication. Nationwide Insurance in Columbus, Ohio, for instance, has found a way to get its business analysts and software coders to, if not bond, at least interact.

"They sit next to one another," said Reece. "The business analysts have to learn the logic that goes into the code, and the coders have to learn the business."

This interaction helps break down departmental barriers and build relationships.

"The business analysts don't just throw their requirements over the wall at the start of the project, and come back six or eight months later and expect an end project," Reece added. "They're very integrated throughout the project."

"By empowering our organization to build trusted relationships, we have accelerated our big data and analytics objectives," said Wes Hunt, VP of analytics at Nationwide Insurance, in a statement. "We break down trust barriers by an increased focus on education and personal interaction, so everyone understands and interprets data similarly."

The IBM survey also found two other key issues hampering big data projects: The lack of skills to analyze and interpret data; and the dearth of C-suite executives acting as "lead advocates" for analytics efforts.

"It's interesting that only one quarter of CEOs or COOs are the lead advocates for the use of analytic insights, even though they realize that innovation and revenue growth is the chief value of applying analytics," Reece told InformationWeek. "So we need to get more senior leadership advocating for the use of analytics."

Making decisions based on flashy macro trends while ignoring "little data" fundamentals is a recipe for failure. Also in the new, all-digital Blinded By Big Data issue of InformationWeek: How Coke Bottling's CIO manages mobile strategy. (Free registration required.)