• by Karl Greenberg , August 5, 2013

How big is Big Data? By 2017, per global analytics firm Teradata, three billion people will create eight zettabytes of data -- slightly more than will fit on your MacBook Pro. Companies will boost market analytics spend by 60%, and CMOs will outspend CIOs on tech by then. The Dayton, Ohio-based company says that marketers are using -- or are hoping to use -- Big Data to improve efficiency, cross-channel integration and metrics and outcome ROI.

Are they anywhere near this? Well, about half of respondents to the firm’s Data-Driven Marketing Survey 2013, a worldwide poll of over 2,200 marketers, say their companies aren't using data correctly. Fewer than 10% of respondents said they are systematic in how they use data. But nearly three-quarters of them said they plan to ramp up analysis platforms in the next two years. 

Forty-two percent of survey respondents blame their bad decisions on a lack of process around data analysis. If an integrated, single view of what their consumers are doing is the goal, not many -- fewer than 20% -- said they have that capability, and three-quarters are at a loss to figure out ROI because they don't have the right level of integration. 

The study suggests -- surprisingly, given the dissolution of traditional media channels -- that marketers generally haven't gotten rid of departmental silos. Sixty-five percent of respondents said that that is precisely what makes it impossible to obtain any kind of a holistic view of a campaign. For example, three-quarters of them assert that their IT departments and marketing departments are not working together strategically.    

When it comes to priorities, the study finds that companies with high use of data don't put a priority on using it to achieve customer experience improvements, channel attribution or to develop big data skills. Among those marketers who put a high priority on improving efficiency, cross-channel integration, customer loyalty and identifying cross-sell, the majority concede they are not using enough data to achieve it.