Not a week goes by when I don’t receive at least a dozen mes­sages about Big Data. Since I focus on the Telco indus­try, I receive many invi­ta­tions to webi­nars and events promis­ing new insights into the now age-old ques­tion: “How can oper­a­tors dif­fer­en­ti­ate them­selves from the com­pe­ti­tion by har­ness­ing the power of Big Data?”

Don’t get me wrong, the topic is of real inter­est. A March 2014 analy­sis from McK­in­sey & Co (titled “Big Data in Tele­coms: How to Cap­ture Value from Cus­tomer Infor­ma­tion”) esti­mates the poten­tial oppor­tu­nity stem­ming from Big Data and advanced ana­lyt­ics is on the upwards of $300 bil­lion. The out­puts of these Big Data efforts sup­port improve­ments in core busi­ness as well as new sources of rev­enue growth. Oper­a­tors have a dis­tinct advan­tage over other indus­tries in terms of the trea­sure trove of data that pass through their net­works. Hence, they have the most to gain by pur­su­ing Big Data initiatives.


Unfor­tu­nately, as is often the case with Big Data, the path to suc­cess is fraught with com­plex­i­ties (i.e. lim­i­ta­tions on stor­ing loca­tion data or min­ing inter­nal data stores for com­mer­cial pur­poses) and exam­ples of Big Data efforts gone wrong. (Remem­ber when Google Flu Trends sub­stan­tially over­es­ti­mated instances of the flu in 2012 and missed the 2009 Swine Flu pandemic?)

I recently read a piece by Booz & Co. that sug­gested that the top-down method to solv­ing Big Data ques­tions may not be the best approach for at least two rea­sons: “[First] the busi­ness prob­lem often exceeds the capac­ity of the avail­able data to solve it, and sec­ond, the process of gath­er­ing the right data to help solve the prob­lem is poorly under­stood by many com­pa­nies.” The arti­cle sug­gested tam­ing data with a bottom-up approach. I’ve seen bottom-up approaches yield tan­gi­ble ben­e­fits for many orga­ni­za­tions in the Telco indus­try and in other indus­tries as well. The approach also lends itself to agile friendly mar­ket­ing, which has its own benefits.

Step 1. Iden­tify Highly Pre­dic­tive Data Sets

As a first step in prepar­ing your data for bottom-up analy­sis, iden­tify the data sets that are the most pre­dic­tive. Some data are more likely to offer mean­ing­ful cor­re­la­tions and there­fore should be incor­po­rated as the foun­da­tion of robust cus­tomer pro­files and pre­dic­tive mod­els. For exam­ple, a report from Pew Research shows there is a strong cor­re­la­tion between device fea­tures and demo­graph­ics: 25–34-year-old males with house­hold incomes of $75,000+ and edu­ca­tion level ranges from high school edu­ca­tion to some col­lege have a strong pref­er­ence for Android OS.

Usage data can also be a strong indi­ca­tor of future behav­ior; as such, voice and data con­sump­tion are often incor­po­rated as part of audi­ence seg­men­ta­tion schemas. Another impor­tant activ­ity is the map­ping of behav­ioral, or implicit data, gath­ered through cus­tomer engage­ment with dig­i­tal chan­nels such as Web, mobile, and email. The results of data dis­cov­ery will reveal hid­den rela­tion­ships and pro­vide fur­ther evi­dence that not all data is equal.

Step 2. Run Clus­ter Analy­sis Across Data Sets

After iden­ti­fy­ing the most pre­dic­tive data sets and seg­ments, the next order of busi­ness is to per­form clus­ter analy­sis across the col­lected data sets. Tech­nol­ogy advances over the recent past have greatly eased the once ardu­ous task of iden­ti­fy­ing sta­tis­ti­cally sig­nif­i­cant events. Sophis­ti­cated ana­lyt­ics envi­ron­ments with visu­al­iza­tion capa­bil­i­ties (such as the Adobe Ana­lyt­ics Data Work­bench) are now a foun­da­tional part of mar­ket­ing ecosys­tem that sup­ports cutting-edge mar­ket­ing orga­ni­za­tions. The prac­tice of algo­rith­mic mar­ket­ing, whereby data gen­er­ated insights are exe­cuted in real time, is tak­ing hold and oper­a­tors are sci­en­tif­i­cally man­ag­ing a broad spec­trum of mar­ket­ing issues, such as tar­geted offers based on Next Prod­uct to Buy (NPTB) mod­els and pro­gres­sive offers designed to deflect churn. (See the 2012 McK­in­sey report, “By the Num­bers: Unleash­ing the Power of Algo­rith­mic Marketing.”)

Orga­ni­za­tions that are able to tame their data are able to make deci­sions ahead of the curve and are suc­cess­ful in deliv­er­ing effi­ciency and prof­itabil­ity across the entire value chain:

  • Prod­uct and services
  • Net­work operations
  • Sales and marketing
  • Cus­tomer service

If you would like to read more on this sub­ject, I have explored the topic of Big Data for Tel­cos in more detail in the Adobe whitepa­per “Scale Big Data into Mas­sive Insights.”