• by David Baker , Op-Ed Contributor, May 5, 2014

This quote spurred me to think this past week:   “He uses statistics as a drunken man uses a lamp post -- for support rather than illumination” --  Andrew Lang, a Scottish poet.

While we profess to have data at our fingertips, how do you think of this data? What do you have to compile yourself, and what data is out there that you struggle to use in any meaningful way? 

I try to simplify this into a few categories I like to call the Four Ps -- not to be confused with the marketing mix:

Personal data:  This is the publicly accessible information defining who you are:  name, address, job, age, home owner/renter, presence of children, marital status, ethnicity, etc.  This information is typically gathered through direct transactional or external sources and can be very expensive to compile.  

Profile:   This is how we look at classifications:  education, religion, socioeconomics, religious affiliations, along with many derived elements that project more depth to your lifestyle, lifestage, and personal/family characteristics. 

Preferential:  This type of data can be self-disclosed or implied from some online behavior, and can be very valuable -- but also quite misleading if not used properly. While you may have a high propensity to buy a hotdog at a baseball game, the reality of you doing so is not really that relevant if you never attend a baseball game. This, to me is some of the coolest data to play with, if you wake up in the morning wondering what a billion people are doing.  If applied correctly to a channel, this can enhance future trends of engagement, frequency of engagement and building episodic programs that trend with the consumers.   

Performance: This is exactly what you think: purchases, buying patterns and online behaviors. This is where most email geeks profess to have strengths, but can also be seen as trying to optimize the room temperature.  (just not that important unless there’s a heatwave or freeze).       Email response data doesn’t really inform much unless trended over time or applied to something that happened in-market.   With all the social consumer shifts, mobile/device shifts and place shifting, this data is transient and very hard to visualize, but much easier to automate into programs.

One could meander for days on all the data sets, and debate would continue for weeks.  Fortunately I get to see more than most folks in terms of  what’s available, and what is most updated across industries. Here are more recommendations:

Personalization shouldn’t be thought of as hit or miss, just contextually relevant (gender, relevant interests).

Timing and device are the ground zero for success.     Fact is, it is very hard to buy on a 5” device, and most trends suggest the buying and browsing patterns are following the tablet vs. the mobile device.   If a transaction is your goal, follow the tablet and the customer.  If viewership and monetizing impressions and one click behaviors is your goal, optimize “audiences” (both anonymous and non-anonymous). 

Always prioritize the information available, figure out what to visualize, what to put in the sand box and what to really challenge your quant geeks to make sense of.