For those of us involved in the col­lec­tion and analy­sis of data, we sim­ply can’t avoid the topic of Big Data. It now has the dubi­ous honor of being the most con­fus­ing buzz­word of the decade—and for good rea­son. It’s a label that is lib­er­ally applied across the tech­nol­ogy land­scape. From indus­try thought lead­ers try­ing to one-up each other to com­pa­nies try­ing to make their prod­uct releases more appeal­ing, Big Data has popped up every­where, and one can eas­ily feel over­whelmed try­ing to find its promised oasis of unpar­al­leled insight in a desert of attrac­tive look­ing mirages.

This is the first in a series of post­ings where I’ll be talk­ing about some of the illu­sions that I see sur­round­ing Big Data as it relates to dig­i­tal mar­ket­ing. My intent is to pro­mote the real­ity that any orga­ni­za­tion can take advan­tage of Big Data and that it doesn’t take mil­lions of dol­lars or mov­ing moun­tains to get started.

The Big­ger Picture

First, a real­ity of Big Data is that it can give you a high-resolution view of your cus­tomers. The illu­sion is that sim­ply hav­ing more data will some­how mag­i­cally accom­plish this.

Doing the same analy­ses on a deeper set of data will def­i­nitely increase accu­racy as you move from ana­lyz­ing sam­ples to pop­u­la­tions (which is fan­tas­tic!), but with­out tying in addi­tional data sets or doing dif­fer­ent types of analy­sis (addressed in my next post), new insights will be limited.

With Big Data, it is often assumed that we need to go deeper to gather all the infor­ma­tion we could pos­si­bly need. In fact, we should be going both broader and deeper, mean­ing we need to inte­grate our data sources together.  As data con­tin­ues to mul­ti­ply, more data will be avail­able tomor­row and from more sources than today, and most of it will be unstruc­tured. Who knows what is going to be dig­i­tized and datafied tomor­row? Our cus­tomers are inter­act­ing through mul­ti­ple chan­nels includ­ing mobile, video, web­sites, and point of sale con­tacts daily. Cap­tur­ing this unstruc­tured data set and join­ing it with another data set to see what shakes out—that is when we begin to get to the real­ity of Big Data.

Let me illus­trate: I had an expe­ri­ence this past sum­mer that was a prime of exam­ple of a non­pro­duc­tive use of Big Data. I bought a set of golf clubs from a national chain sport­ing goods store. I had a great in-store expe­ri­ence and even signed up for the rewards pro­gram. How­ever, a few days later, I received an email offer for 25 per­cent off the dri­ver I had just pur­chased. Whoops. All that equity the brand built with me started going down the drain. If their data col­lec­tion was fine-tuned and inte­grated, they would have known that I already pur­chased the club and would have sent me a notice of sav­ings on shoes or some other acces­sory. Instead, the notice served more as a slap in the face or “gotcha” than as a prof­itable mar­ket­ing event.

I’m sure that each team involved prob­a­bly thought they hit a home run. The demand-generation team that sent the email out prob­a­bly saw the per­cent­age of open emails go up. The team in charge of in-store point of sales saw another per­son sign up for the rewards pro­gram. The per­son­al­iza­tion team sent a golf-related email to some­one who has a golf inter­est. Sep­a­rately, the team that sent the email, the ana­lyt­ics team, and the per­son­al­iza­tion team were all under the illu­sion of suc­cess, high-five-ing one another in their indi­vid­ual silos. Was it suc­cess­ful? I have yet to take them up on any offers.

The dis­con­nect is obvi­ous when we peer in from the out­side. For what­ever rea­son (tech­ni­cal, orga­ni­za­tional, polit­i­cal, etc.), these teams are not com­mu­ni­cat­ing and con­nect­ing their data, which results in lost opportunities.

This is what I’m talk­ing about by mak­ing Big Data broad and not just deep. Com­plete, inte­grated data sets don’t have to be mas­sive to real­ize the promise of Big Data for busi­ness, which is know­ing your cus­tomers well enough to pro­vide valu­able expe­ri­ences that keep them com­ing back again and again.

At Adobe, we are get­ting a higher-resolution view of our cus­tomers because we are con­nect­ing data sets and gath­er­ing more data from wider, out­side sources. For exam­ple, Adobe doesn’t own Face­book, but peo­ple post a lot of infor­ma­tion about Adobe there. It is impor­tant that we cap­ture that data and get a closer look at it in rela­tion to all the first-party data we col­lect. As we layer other data (social, email, adver­tis­ing, etc.) on top of our Web data, the pic­ture of our cus­tomers becomes clearer and new insights are revealed.

Real­iz­ing the promise of Big Data can start with one con­nec­tion. You don’t need to boil the ocean. For instance, most Web ana­lytic providers have out-of-the box inte­gra­tions with many email providers. It’s a quick win for new insight and can help pave the way for orga­ni­za­tional buy-off as you un-silo more com­pli­cated data inte­gra­tion.

A word of cau­tion here though: keep in mind that with more data comes more types of analy­sis and more room for “data fit­ting.” One of the car­di­nal sins of data analy­sis is mis­us­ing data to jus­tify a course of action based on an agenda. Good data gov­er­nance will keep this to a minimum.

Striv­ing for Completeness

Let’s talk about data com­plete­ness a lit­tle more. A lot of orga­ni­za­tions that try to get a han­dle on Big Data will start to col­lect every bit and byte under the sun and then sam­ple their data sets, cre­at­ing more prob­lems. The key is to get closer to the pop­u­la­tion with the goal of com­plete­ness, not just a big silo of data. Full data sets pro­vide the free­dom to explore, much like a high-resolution photo, but this is where sam­pling can cause it all to fall apart, like turn­ing that high-res photo into a small JPEG file.

We can be more con­fi­dent with our pre­dic­tions with more sam­ple (or com­plete) data than with smaller sets. Frankly, that’s the whole point of cap­tur­ing data: to run ana­lyt­ics on it and pre­dict what offer or action is going to make our cus­tomers pull the prover­bial trig­ger on a purchase.

Big Ana­lyt­ics

Cut­ting through the hype and get­ting to the real­ity of the use of Big Data is nec­es­sary to improve your cus­tomer engage­ment strate­gies. The mis­con­cep­tion regard­ing Big Data is that once you’ve obtained it, your busi­ness will nat­u­rally improve its bot­tom line based on the infor­ma­tion you’ve received. The con­cept of data com­plete­ness, as well as being con­fi­dent in our cus­tomer pre­dic­tions using data, must be aug­mented by inte­grat­ing data sets. Only then can we use ana­lyt­ics to make sense of the data we want to take action on.

In Part 2, I will address Big Data ana­lyt­ics. I sub­mit that Big Data doesn’t make any sense until you run ana­lyt­ics on it. Big Data … or “Big Ana­lyt­ics”? Where does value reside? What say you?

Naomi Civins
Naomi Civins

The distinction between breadth and depth is so crucial. We get the most valuable insights when we connect different sets of data, finding relationships and correlations and getting into the real detail.