When it comes to Big Data, it is hard to overstate the importance of having a clear business strategy. Without this strategy, it is tough to attain a return on investment.
Joint research by Capgemini and Informatica found that less than a third of Big Data projects are considered profitable. Another 45 percent are breaking even, 12 percent are losing money, and 12 percent cannot yet determine whether their Big Data projects are profitable.
The numbers are not that surprising, said Steve Jones, Global VP of Big Data for Capgemini, given that many Big Data projects were undertaken as projects that aimed to tackle the technical learning curve and thus focused on goals such as how to build a Big Data cluster. "They were not even looking for ROI; their journey was about proving and understanding, not about ROI," he said.
Companies that take a more strategic approach to Big Data are experiencing better returns, the report titled "The Big Data Payoff: Turning Big Data into Business Value" seems to indicate.
For instance, 49 percent of those in the midst of ongoing, enterprise-wide Big Data projects say they are profitable, vs. 25 percent of organizations with ongoing projects in individual departments or groups. Tellingly, no companies running limited projects in just some departments or groups were profitable.
Big Data's Geographic Advantage
European companies are more likely than their U.S. counterparts to take an enterprise-wide approach to Big Data, the research found. Thirty-six percent in Europe do so, vs. 21 percent in the U.S.
Because Europeans were more conservative in adopting Big Data, they have had time to observe its evolution. Capgemini is now seeing some U.S. companies undertaking a refresh of their legacy Big Data environments, Jones said, adding that it seems odd to apply the term "legacy" to a concept as relatively new as Big Data.
"We are working with several large banks with departmental solutions who now want to implement Big Data across the bank," Jones said. "Companies without 'legacy' Big Data can start from a more strategic place."
In fact, Jones wrote in a blog post, it is possible this trend is giving Europe a second-mover advantage in Big Data. "Europe has woken up to Big Data and is going at it in a measured and industrialized way. The U.S. is dealing with the point and departmental solutions and struggling to get the sort of consistent value that a strategic model would deliver," he said in the post.
According to the research, 30 percent of respondents plan to accelerate or expand current Big Data initiatives to new departments or locations in the next 12 months. Again, this is a more popular goal with Europeans (35 percent) vs. U.S. companies (23 percent).
Business Involvement in Big Data
The research also found that 33 percent of profitable Big Data initiatives are led by COOs. In contrast, just 14 percent of unprofitable Big Data projects have COO leadership.
Ownership appears to be shifting to business roles such as CMO and COO as companies determine more profitable strategies for Big Data such as using it for operational analytics, Jones said, noting, "Business leaders want to lead initiatives that deliver business value."
Executive buy-in is another important success factor. Forty-nine percent of companies with support from the CEO/executive team say their Big Data initiatives are profitable, vs. 6 percent of respondents with executive teams that have a dim view of Big Data's potential to achieve benefits.
The top-ranked priority for operationalizing Big Data is improving data quality and data governance, cited by 77 percent of respondents. Other top priorities include improving data security, mentioned by 74 percent, and standardizing and improving consistency across the organization (69 percent).
These types of priorities show that "Big Data has grown up," Jones said. "A lot of concerns and challenges are the same as what you'd have for an SAP implementation. That shows that Big Data is becoming better integrated into the business."
Organizations that have already achieved profitability are more likely than others to have already made progress in these areas. For instance, three-quarters of profitable companies say they’ve made excellent or very good progress in improving data quality and data governance, compared to half of overall respondents.
Better Data Governance
Big Data in some ways is "forcing companies to do good governance," Jones said. "You cannot do bad governance when you include IoT data, for instance. You cannot say, 'I disagree with what that airplane is putting out as data.'"
Effective data governance begins with IT understanding data and information needs from a business perspective. "The problem has been that IT often tries to enforce an IT governance on the business; they are surprised when business isn't interested," Jones said, offering the example of a IT organization that tried to impose a standard invoice on a company that produced a diverse array of products, such as consumer products and parts for satellites.
Business and IT organizations must work together to improve data governance, he said. Conversations between the two can result in some surprising insights. A client company that makes turbines was getting IoT data with misreads. "If the turbines are spinning at 10,000 rpm and then it says they are spinning at 100 rpm, that is just not possible," Jones said.
IT wanted to remove that data in the aim of preserving data quality. However, a business person pointed out it would make it easier for the company to improve its sensors if it knew more about what caused misreads.
"One of the advantages of Hadoop is that you can load data in raw and add appropriate qualities later," Jones said. "Sometimes what we think is bad data from an IT perspective turns out to be good data from a business perspective."
One Capgemini client has a coffee bar where business people come into regular contact with data professionals. Both sides are pleased when business people find that at least some of the questions they want to address can be answered quickly using Big Data technologies, Jones said.
"With a data warehouse, it takes six months to deliver anything," he said. "But with Big Data and new methodologies like DevOps IT can deliver at least some things fast."
Ann All is the editor of Enterprise Apps Today and eSecurity Planet. She has covered business and technology for more than a decade, writing about everything from business intelligence to virtualization.