Does too much data kill data?

Analytics

Today’s arti­cle is a response to the recent pub­li­ca­tion in the French mag­a­zine “L’Usine Dig­i­tale” of the inter­view of the leader of one of the most impor­tant retail chains in France, about his vision of dig­i­tal. Asked about the need to cap­ture and analyse the data for retail­ers, he responds, “With Big Data, we are in a bub­ble that, when it will explode, will make con­sid­er­able dam­age. Too much infor­ma­tion kills infor­ma­tion. ”

Beyond the ref­er­ence to the Amer­i­can econ­o­mist Arthur Laf­fer and his for­mu­la “Too much tax kills tax ‘, which means both every­thing and noth­ing, and that every­one tends to use in its own way, this major leader seems to think that we are trapped under a huge mass of data that we don’t have the capac­i­ty to process, and that this data„ there­fore doesn’t nec­es­sar­i­ly help us to make clever deci­sions.
These state­ments made me react because my vision is dif­fer­ent: data exploita­tion and what is called Big Data are issues with both a human and a tech­no­log­i­cal dimen­sion. An over­flow of infor­ma­tion blocks the human capac­i­ty to process this infor­ma­tion, but in no case those of machines, which have a pro­cess­ing capac­i­ty far supe­ri­or to ours. The chal­lenge here is for me one of capac­i­ty, not exper­tise.

A human dimen­sion

This is to assess the human capac­i­ty to exploit these thou­sands of data col­lect­ed every day. From what hap­pens in stores to the behav­iour of a client on the mobile app of the brand, or the way a con­sumer acts in the park­ing lot of a store or in the aisles of the super­mar­ket, the aver­age bas­ket or the types of prod­ucts they buy … Mar­keters now have bil­lions of data avail­able to them: yet, it is human­ly impos­si­ble to process these bil­lions of data, hence the impor­tance of the tech­no­log­i­cal dimen­sion and the use of suit­able tools.

A tech­no­log­i­cal dimen­sion

In real­i­ty, the chal­lenge today is no longer to cap­ture infor­ma­tion to get some infor­ma­tion, but rather to retrieve data cor­re­spond­ing to spe­cif­ic use cas­es, for which we need to offer a cus­tomized mar­ket­ing response.

This is par­tic­u­lar­ly what does Adobe Ana­lyt­ics, through their Con­tri­bu­tion Analy­sis Mod­ule: it is based on data col­lect­ed by Adobe Ana­lyt­ics, in order to process infor­ma­tion a human brain would not be able to get (or only after a very long peri­od of time), and mea­sure the impact of a par­tic­u­lar event on a result.

Let’s take the exam­ple (real but that unfor­tu­nate­ly can­not be named) of a B-to-B trad­er, who sud­den­ly real­izes a 81% increase in orders. This mer­chant has assigned a team of five data sci­en­tists to try to iden­ti­fy the rea­sons of this increase for a whole week­end: unfor­tu­nate­ly, they were able to analyse only 5 out of 300 poten­tial­ly rel­e­vant dimen­sions. How­ev­er, it took only 30 sec­onds for the tool to real­ize that this peak in orders was relat­ed to fraud­u­lent dis­count vouch­ers. This find­ing then enabled man­agers to delete these dis­count vouch­ers, and to can­cel the fraud­u­lent orders. In this case, solv­ing the prob­lem was pos­si­ble only because of the tech­no­log­i­cal capa­bil­i­ties of the mod­ule, due to the vol­ume of data to process.

Anoth­er inter­est­ing exam­ple: a com­pa­ny belong­ing to the trav­el indus­try had real­ized that it had a dai­ly a sig­nif­i­cant dai­ly Increase Rev­enue of $ 1.7 mil­lion. After analyse from the tool, it turned out that the most prof­itable cam­paign for this client had been dis­abled due to a mis­in­ter­pre­ta­tion of the data from the Ana­lyt­ics team. Sim­ply reac­ti­vat­ing this cam­paign has enabled this cus­tomer to stop los­ing upwards of $1.7M/day.

It’s in this sense that we can­not say that “too much infor­ma­tion kills infor­ma­tion”: the machine has a pro­cess­ing capac­i­ty far supe­ri­or to ours, but we need human for the added val­ue that it rep­re­sents and its deci­sion-mak­ing pow­er. The two are indis­pens­able and the two, com­bined, allow com­pa­nies to make deci­sions incred­i­bly more prof­itable than they were a few years ago…

What about you, what is your opin­ion on the use of Big Data? Feel free to con­tin­ue the dis­cus­sion and share your views on the sub­ject in the com­ments sec­tion!


Analytics
Olivier Binisti

Posted on 10-12-2015


Join the discussion