5 lessons from Face­book on an­a­lyt­ics suc­cess

Ken Rudin, Face­book’s head of an­a­lyt­ics, shares five strate­gies the so­cial net­work uses to im­prove adop­tion and turn busi­ness in­sights into ac­tion

InformationWeek - - Contents -

Ken Rudin, Face­book’s head of an­a­lyt­ics, shares five strate­gies the so­cial net­work uses to im­prove adop­tion and turn busi­ness in­sights into ac­tion

Busi­ness an­a­lyt­ics teams can’t just gen­er­ate in­sights and leave it up to the busi­ness to put them to use, says Ken Rudin, head of an­a­lyt­ics at Face­book. The goal should be to make an im­pact on the busi­ness, and that means tak­ing own­er­ship of the chal­lenge of turn­ing in­sights into ac­tion.

So what do you do to make sure the an­a­lyt­ics team’s ef­forts to de­tect data cor­re­la­tions, spot cus­tomer be­hav­ior, and un­cover pos­si­ble im­prove­ments and ef­fi­cien­cies don’t go to waste? In a re­cent in­ter­view with In­for­ma­tion­Week, Rudin shared five sug­ges­tions for driv­ing an­a­lyt­ics adop­tion:

MAKE IT UN­DER­STAND­ABLE

It’s one thing to give sales­peo­ple lists of prospects. It’s an­other thing to ex­plain that you have al­go­rithms that an­a­lyze buy­ing pat­terns and pur­chase his­tory to spot ripe prospects for par­tic­u­lar upselling or cross-sell­ing of­fers. It’s even bet­ter if you pro­vide an in­ter­face that lets users play with buy­ing vari­ables and ex­plore prospect data on their own.

“If you don’t un­der­stand what’s be­hind the anal­y­sis, it’s more likely you’ll fall back on in­tu­ition,” says Rudin.

MAKE IT VIS­UAL

“The more fric­tion you add, the fewer people will use the data, and the harder it will be to make an im­pact. We’re us­ing a va­ri­ety of vi­su­al­iza­tion tech­nolo­gies so we can move away from dump­ing ta­ble ex­tracts into Ex­cel.”

MAKE IT DIS­COV­ER­ABLE

Es­tab­lish one place where people go to find analy­ses and cer­ti­fied data or­ga­nized by topic. It can be as sim­ple as a por­tal with a set of pointers, but it must be eas­ily ac­ces­si­ble and com­pre­hen­sive.

USE A HY­BRID OR­GA­NI­ZA­TIONAL MODEL

Face­book’s data an­a­lysts col­lab­o­rate with one an­other and get strate­gic di­rec­tion from Rudin in a cen­tral­ized way, but they are em­bed­ded within busi­ness units, so the op­er­a­tional work is de­cen­tral­ized. This ap­proach en­sures a level of cen­tral­ized con­trol, so data an­a­lysts can learn from their peers and don’t du­pli­cate ef­forts. At the same time, day-to-day work is in tune with busi­ness needs and ben­e­fits from the di­ver­sity of hav­ing data an­a­lysts work­ing along­side busi­ness unit lead­ers, project lead­ers, de­sign­ers, and en­gi­neers. The roles within your team may dif­fer, but the idea is to guar­an­tee a bal­ance of per­spec­tives.

TRAIN EM­PLOY­EES

Face­book holds in­ten­sive, two-week-long Data Camp train­ing ses­sions for an­a­lysts and line-of-busi­ness em­ploy­ees. One goal is to train people on a va­ri­ety of data anal­y­sis tools. More im­por­tant, how­ever, are ses­sions de­signed to spark think­ing about how to solve the re­al­world busi­ness prob­lems sub­mit­ted by Face­book’s busi­ness units.

“We have a run­ning collection of prob­lems that busi­ness units are work­ing on, and we ask the Data Camp par­tic­i­pants — an­a­lysts, project man­agers, de­sign­ers, en­gi­neers, people from fi­nance — to think through the prob­lem.” In some cases, Data Camp work has con­trib­uted to solv­ing re­al­world prob­lems.

The train­ing ses­sion your or­ga­ni­za­tion uses may not be as long as two weeks, but Rudin says it’s im­por­tant to make it a full-time com­mit­ment with­out forc­ing em­ploy­ees to keep up with e-mail and their usual re­spon­si­bil­i­ties. “We give them a lot of work, and they’re ex­pected to come up with an­swers and in­sights that re­ally lead to an im­pact on the busi­ness.”

As for the im­per­a­tive to make an im­pact, Face­book en­cour­ages teams to look be­yond ask­ing ques­tions; you have to think through what you would do dif­fer­ently if you could an­swer the ques­tion. “If the team doesn’t know what it might do dif­fer­ently, I’m go­ing to choose a dif­fer­ent prob­lem.”

Ken Rudin

Head-An­a­lyt­ics, Face­book

1

2

4

5

3

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.