Guidelines for Developing a Big Data Strategy

Where should we begin?

Yesterday the DMA was bookended by two well-attended sessions on Big Data.   In the morning Bruce Biegel of Winterberry Group covered the flow of ad spend and the trends they are seeing.  In afternoon Yahoo and Acxiom collaborated on a discussion of linking display ads to offline sales.  In combination they provide some structure for developing a big data strategy for marketing.

An organization must address the following areas.

1. Use Cases: defining your area of focus within the marketing landscape helps not only navigate and prioritize the business requirements but also highlights organizational alignment challenges.   Calculating ROI via cross-channel attribution in order to make media mix decisions to reach audiences based on their behavior in real-time while staying compliant needs a fresh perspective. 
2. Acceptable Use:  the most critical step is to articulate the boundaries of permission, tolerance, risk and privacy by each combination of channel and audience.   Three distinct lenses need to be applied to each area - consumer, marketer, and publisher.   For some use cases safe harbor functions need to be created and probably from scratch since firms aren't typically organized across such large gulfs. Communication of your intentions must be transparent along with sufficient choice mechanisms given to each of the other parties in the system.  
3. Integration: the alignment of anonymous behavior with personally identifiable information requires a new way of thinking about data structures, management and analysis that is neither digital nor direct to consumer focused.   The current key to linking sales and activity is the email address since it satisfies both requirements marketers have - recognition and activation.   The data architecture must start from a much higher level than any delivery mechanisms used today as well as cover new forms of content that typically were beyond the remit of the data team.

For example, while the idea of replicating direct mail online sounds simple the governance and integration issues are vastly different.

Still a good time to be a marketing technologist.