Attack of the Killer Data Scientists: How to Prepare for the Invasion

by | Sep 7, 2017 | ana­lyt­ics

This blog is inspired from a top­ic I pre­sent­ed this year at the Observe­Point User Groups in New York and Boston. The audi­ence thought it was pret­ty cool and I hope you do to.

The last five years of my career have been built on a shift in the way we “do” dig­i­tal analytics.

I entered this space back when most com­pa­nies had two groups of ana­lyt­ics peo­ple: one group of tech­ni­cal peo­ple who installed ana­lyt­ics tools (usu­al­ly a web dev team), and the mar­keters who used said tools (usu­al­ly me). 

I was lead­ing the dig­i­tal ana­lyt­ics team at the Amer­i­can Can­cer Soci­ety in 2012 when we became ear­ly adopters of a tag man­age­ment tool called Satel­lite (devel­oped by Search Dis­cov­ery, now Adobe DTM).  It was love at first sight, and I quick­ly became the world’s biggest evan­ge­list of this tool. This is not a shame­less plug, I real­ly was pas­sion­ate about it and told any­one who would lis­ten about how it would rev­o­lu­tion­ize the way we do dig­i­tal ana­lyt­ics. Even­tu­al­ly, I left my job at ACS to join Search Dis­cov­ery, where I’ve spent the last four years help­ing oth­er dig­i­tal ana­lyt­ics teams har­ness the pow­er of tag management.

Tag man­age­ment changed a lot about how we do our jobs, and more impor­tant­ly the types of skills we need to be suc­cess­ful. Ana­lyt­ics prac­ti­tion­ers pro­vide the same val­ue today that we did before tag man­age­ment, but the way we do our jobs has shift­ed dra­mat­i­cal­ly.  I believe that data sci­ence is bring­ing the next shift. 

The Status Quo

What Makes A Suc­cess­ful Dig­i­tal Ana­lyt­ics Team

If you read enough job descrip­tions online for a Dig­i­tal Ana­lyst, you’ll notice a theme: ana­lysts are expect­ed to solve busi­ness prob­lems.  Here are a few exam­ples I pulled this week:

  • Guide the use of data mea­sure­ment to dri­ve the strate­gic roadmap”
  • Busi­ness-savvy and action-ori­ent­ed strate­gic leader”
  • Self-starter with out­stand­ing ana­lyt­ic skills, strong busi­ness acu­men and the abil­i­ty to syn­the­size data into mean­ing­ful insights”

And I don’t dis­agree. In fact, I sum­ma­rize the sen­ti­ment like this:

Dig­i­tal ana­lyt­ics teams pro­vide lead­er­ship on inter­pret­ing data to advance the business.”

I’ve con­sult­ed a lot of ana­lyt­ics teams over the years, and although every com­pa­ny is dif­fer­ent, they all do four things on a dai­ly basis to achieve this goal: 

  1. Col­lect data — Own the tag man­ag­er, or work close­ly with the dev team.
  2. Oper­a­tional­ize data — Dis­trib­ute base­line infor­ma­tion through dash­boards and reports.
  3. Reac­tive­ly inter­pret data — React to busi­ness ques­tions, such as “Why was rev­enue down last month?”
  4. Proac­tive­ly inter­pret data — Iden­ti­fy oppor­tu­ni­ties to dri­ve new busi­ness, such as “A blog­ger drove a ton of traf­fic last week, we should reach out to them about an affil­i­ate oppor­tu­ni­ty.”  Or, “This list of prod­ucts have a poor con­ver­sion rate from users who view the prod­uct detail page, so we should update that con­tent to make it more appealing.”

Just about every task that we con­sid­er to be dig­i­tal ana­lyt­ics falls into one of these four cat­e­gories (I would even stick A/B test­ing into #3 or #4 since it involves mak­ing a hypoth­e­sis, inter­pret­ing data, etc).

Get­ting Good Data

If our goal is to pro­vide lead­er­ship on inter­pret­ing data to advance the busi­ness, then #4 is where we real­ly show our val­ue.  Unfor­tu­nate­ly, most of my clients strug­gle to find time for proac­tive inter­pre­ta­tion because they’re con­sumed by try­ing to make sure they’re col­lect­ing the right data or are just churn­ing out KPIs to the business.

Items #1 and #2 are just about keep­ing the lights on. Tag­ging web­sites and mobile appli­ca­tions are thank­less tasks that are more com­plex than most peo­ple real­ize. No one is impressed when they are done well, but they get very upset when there is a prob­lem. After all, no job descrip­tion says, “seek­ing some­one to work through the week­end to fig­ure out why rev­enue in Adobe Ana­lyt­ics dif­fers from rev­enue in our account­ing tool”!

Change is Coming

The world we oper­ate in today requires us to dis­pro­por­tion­ate­ly invest our time in data col­lec­tion and oth­er tasks that offer lit­tle busi­ness val­ue. But I have a prediction: 

The prob­lem of col­lect­ing data is going to be resolved

It won’t be tomor­row and it won’t hap­pen overnight, but tech­nol­o­gy will even­tu­al­ly reduce the amount of effort required to col­lect and main­tain good data. We’ve come a long way from col­lect­ing serv­er logs to the way we work tag man­agers today, and we are mov­ing quick­ly in the direc­tion of stan­dard­iz­ing and automat­ing data col­lec­tion. When the prob­lem of col­lect­ing data from web­sites and mobile appli­ca­tions is resolved, the demand for high­ly skilled imple­men­ta­tion spe­cial­ists will inevitably decline.  

So, those of us who have built our careers on this demand must ask our­selves: how will our pri­or­i­ties shift as the chal­lenge of get­ting good data is resolved, and how do we pre­pare our­selves for this change? The answer is through data science.

How data sci­ence skills will make future ana­lyt­ics teams successful 

Imag­ine a world where ana­lysts have excel­lent data at their fin­ger­tips, tak­ing data col­lec­tion and data oper­a­tional­iza­tion off their plates. and they still have the same respon­si­bil­i­ty: to pro­vide lead­er­ship on inter­pret­ing data to advance the busi­ness.  What would their pri­or­i­ties be, and what skills would they need to do them well?

To answer that ques­tion, I look to the work being done today by peo­ple who focus exclu­sive­ly on inter­pret­ing data, whom I am broad­ly defin­ing here as data sci­en­tists.  The def­i­n­i­tion of data sci­ence is under debate, and not every aspect of data sci­ence applies to dig­i­tal ana­lyt­ics, but I find the fol­low­ing from the Data Sci­ence Ini­tia­tive at NYU to be helpful:

One way to con­sid­er data sci­ence is as an evo­lu­tion­ary step in inter­dis­ci­pli­nary fields like busi­ness analy­sis that incor­po­rate com­put­er sci­ence, mod­el­ing, sta­tis­tics, ana­lyt­ics, and math­e­mat­ics.” (http://datascience.nyu.edu/what-is-data-science/)

Con­cur­rent­ly, while we have been mak­ing progress to solve the prob­lem of col­lect­ing good data, the his­toric bar­ri­ers to entry into data sci­ence have been quick­ly dis­man­tled by the increase in free train­ing mate­ri­als and the decrease in cost for cloud computing/storage.  If we con­sid­er that we will be spend­ing less of our ener­gy on data col­lec­tion in the next 5 or 10 years, and think of data sci­ence as the next evo­lu­tion­ary step in dig­i­tal ana­lyt­ics (which is becom­ing increas­ing­ly attain­able), it is help­ful to explore how com­put­er sci­ence, mod­el­ing, sta­tis­tics, and math­e­mat­ics are already being applied to dig­i­tal ana­lyt­ics prob­lems today. 

Here are three exam­ples that are occur­ring on the fringes of our indus­try today, but which I pre­dict will be com­mon­place in the next 5–10 years:

  1. Anom­aly detec­tion: Alert the busi­ness to unusu­al activ­i­ty.  This is being used today to mon­i­tor user behav­ior, as well as to iden­ti­fy poten­tial errors with­in your track­ing code.
  2. Cal­cu­lat­ing lift: Under­stand the return on invest­ment we gen­er­ate with the projects we tack­le by mea­sur­ing how a met­ric per­forms rel­a­tive to its fore­cast.  This is being used today to deter­mine if an adver­tis­ing cam­paign was able to increase rev­enue above our fore­cast or to mea­sure if a site redesign effec­tive­ly improved user experience.
  3. User Clus­ter­ing: Assign users into groups that share sim­i­lar char­ac­ter­is­tics and/or behav­iors.  This is being used today to detect the users who are about to make a pur­chase, or users who are inter­est­ed in con­sum­ing cer­tain types of content.

And these are only a few of the ways that data sci­ence capa­bil­i­ties are being applied to com­mon dig­i­tal ana­lyt­ics problems.

This Is How We Prepare

Today, our team at Search Dis­cov­ery is mak­ing sig­nif­i­cant invest­ments to expand our data sci­ence capa­bil­i­ties and solu­tions.  As part of this process, we’ve out­lined three things that any­one work­ing in our indus­try can do to start prepar­ing for the future we’ve out­lined above:

  1. Take advan­tage of free online train­ing tools:  For SDI, this meant ask­ing a few team mem­bers to iden­ti­fy the best resources for devel­op­ing par­tic­u­lar skills, and shar­ing that list with the rest of the team.
  2. Join the com­mu­ni­ty: Pay atten­tion to those mem­bers of the dig­i­tal ana­lyt­ics com­mu­ni­ty who are cur­rent­ly apply­ing data sci­ence tech­niques in our field.  We rec­om­mend join­ing the Mea­sure Slack com­mu­ni­ty of ana­lysts (join.measure.chat) and tak­ing a look at dartistics.com.
  3. Exper­i­ment with ambi­tious projects: Seek out oppor­tu­ni­ties to apply what you’re learn­ing.  This might mean tack­ling a busi­ness prob­lem or sim­ply com­pet­ing in a Kag­gle competition.

We have cho­sen a young, dynam­ic, and con­stant­ly evolv­ing field to work in with dig­i­tal ana­lyt­ics.  I hope you’re excit­ed about that because it means that those of us who are flex­i­ble, pay­ing atten­tion, and eager to learn have the oppor­tu­ni­ty to dri­ve all of us for­ward as a community.

If you’re inter­est­ed in see­ing the full deck from my pre­sen­ta­tion, or if you’d like to learn more about Search Discovery’s approach to dig­i­tal ana­lyt­ics, con­tact me here.