8/12/2019 Invasion of the Data Scientists.

1/8

COVER STORY A n a ly ti c s

nvasion o f th e

Data Sc/entistLeading HR de partme n ts areturn n to talent analytics fa wide ran ge o f s ta ffing issues.

Os a re a t the cen ter of th isdata-drive n tra n sforma tio n

BY STEP HAN IE OVERBY

When General Motors was looking for someone to le d its global talen tand organizational capability group, the $152 billion carmakerclearly wasn't looking for a paper-pushin g adm inistrator. Michael

Arena, w ho took the position 18 months ago, is an engineer by train-ing. He w as a visiting scientist at MIT M edia Lab. He's a Six Sigmailack belt . He's got a Ph.D.

This is not your father's hum an resource s executive.But it is a sign of where the corporate HR function is headed.

Arena is dedicated to the hot field of talent analyticscrunchingdata about employees to get the right people with the right talentin the right place at the right time at the right cost, he says.

Talent management is a soft space. Historically, we haven't beenli le to meas ure definitely the things that w e intuitively believe tobe true, says Arena. But busine sses are ma ndating it. The ageof trust me, this will work is over says Arena. H R is being heldaccountable to deliver business results. And the language of the

busin ess is analytics.The growing importance of sophisticated analytics to HRnot

simply reporting what already exists in an organization but pre-dicting what could or should beis a result of the recognition that

the efficient u se of labor and deploym ent of resources is criticallyimportant to the busin ess resu lts of the company, says Mark Endry,

SEPTEMBER 1 2013 www .cio .co m

8/12/2019 Invasion of the Data Scientists.

2/8

8/12/2019 Invasion of the Data Scientists.

3/8

COVER STORY :: A n a ly ti c s

CIO of Arcadis U.S. He recently spent sixmonths as interim senior vice president ofHR at the $3.3 billion company.

In recent years, enterpr ises have devel-

oped more mature techniques for apply-ing analytics to customer information. They've been able to seewith relativelylittle datahow much they can do andhow powerful the results can be, saysBen Waber, autho r of People Analytics: HowSocial Sensing echnology Will ransform Business and What It ells Us about the Future of ork When you think about what's goingon w ithin companies, you have potentiallybillions of records generated every dayabout each person. They're startin g to seehow valuable and important that data is.

IT must be at the center of the unfold-ing data-driven transformation. Noteveryone has an HR data scientist likeGM. Arena emphasizes the importanceof his partn ership with Bill Houg hton,GM's CIO for global corporate functions.

A big piece is integrationensuring theright systems are connected so we knowwhere to draw the data from, says Arena. IT has to play a role in that.

Indeed, GM's CIO is counting on a newenterprise data warehouseand hiringmore IT professionals with a businessintelligence backgroundto su pport HR'sefforts. Right now the analysis is beingdone by small group of smart people,says CIO Hou ghton. The next step is howdo we make the analytics more availableto the everyday manager or the organiza-tional leadersh ip. We wan t to get this outof the hand s of the rocket scientists andinto the hand s of managers.

CIOs are the key to helping the orga-nization figure out what data matters,

says Terry Sullivan, director of appliedresearch and consulting at office furni-tur e maker Steelcase. Everyone is think-ing about big data a nd collecting all kindsof data to try to figure out how to createsmarter people. CIOs can drive this effort.

IT leaders are uniquely qualified tohelp their corporate counterparts navi-gate the minefield of issues associatedwith these nascent technologies and pro-cessesincluding data quality, system sintegration, security privacy and change

management. The partnership with IT

iscritical, says David Crumley, vice presi-dent of global HR information systems forCoca-Cola Enterp rises.

SEPTEMBER 1, 2013 ww w.cio .com

Talent Analytics 101IT and HR lea ders who are deployingworkforce an alytics sys tems iden tifynine critical success fa ctors

LAY THE FOUNDATIONif possible.

Aim for a s ingle source of HR information.

ACCOUNT FOR IMPERFECTIONS We've got our foundationa i issues,

for sure, but if you wait until it's complete ly perfect, you worn't get an ywhe re,says Michae l Arena, GM s director of global tale nt and organizational capa bil-ity, IT can build reconciliation processes and automated audits to help HR wi thdata issues.

START SMALL Marc Frandosa , CIO of Praxair, began w ith a n an alytics pilotto map the company's high-potential employees. If we had tried to do onebig-ban g workforce a na lytics project, it would neve r have gone an ywhe re,he says. You have to ge t some tract ion In order to get credibility.

TAP INTERNAL EXPERTS Both Franciosa and Arena have taken advan-tage of statistician s and others from the ir corporate R&D groups to develop

theirtalent analytics programs,

SHARE THE LOAD WITH HR Take advantage of HR and ITs complemen -;, tary skills, IT can focus on vendor manage ment s ecuri ty and deployme ntj while HR might manage requirements gathe ring, process s tanda rdization^ and communication.

BRING IN BUSINESS KNOW HOW David Crumley, VP of global HR infor

mation systems for Coca-Cola Enterprise s, works with busines s leade rs fromfunctions such as supply chain, saies and fina nce to determine wha t data wiildrive taien t an alytics.

HIRE EXTERNAL CHANGE MANAGEMENT HELP Typically, HR leadschange manageme nt in an organization, But avoid DIY change man age men tin ana lytics efforts , warn s Mark Endry, CiO of A rcadis U.S., who recen tly spen tsix months as inte rim SVP of HR. Hire e xtern al he lp to guide HR through itsbig changes ,

TAKE ACTION Everyone wan ts to have more data, bu t we have toensure tha t folks know how to use it, says Crumley, who had to do morehand-holding than he initially a nticipated. It's not that an yone is pushingback, but you have to embed the use of the data into the [corporate] DNA.

DEMOCRATIZE THE SYSTEMS For people an aly tics to truly deliver,they nee d to be self-service tools tha t business mana gers an d leaders canuse. Eariy on , we thought the customer [for these to ois] was HR, saysCrumiey, But it's the busines s lea ders tha t control these decisions daily.

- S O ,

8/12/2019 Invasion of the Data Scientists.

4/8

OV R STORY :: n a l y t i c s

Chiquita CIO evin Ledford isexploring predictive analyticsto help the company find rainand retain its bananaeros -experts in growing bananas.

There's a broad array of uses for talentanalytics: screening new hires, figuringout w ho should get promoted, efficientlystaffing new projects, uncovering the char-acteristics of high-performing individuals

or teams, and even predicting who's likely to head out the door. The way I think about it is using data to understand how

people get work done, says Waber, CEO of Sociometric Solu-tions, a management-services firm that was built on his workat MIT Media Lab and that helps companies in one niche ofthe talent analytics field: collecting and analyzing sensor datato improve workforce performance.

Companies have collected employee data for yearsfromsatisfaction surveys to ethnography. But, says Waber, this'next generation of stuff is m oving away from those qualitativeassessment modes into much harder behavioral modes, using

digital data from email or sensors or ERP systems. That givesus radically more powerful information.

Historically, HR used data to report headcount or turnoverinformation. We're so far beyond that now, says Crumley of

Coca-Cola Enterp rises. HR wa nts to expand its capabilities tohelp the business grow. To do so, we need to be able to be moreprecise and surgical about our interventions. That's whereworkforce analytics is hugehelping you determine where

to place your bets.

Layiing the FoundationEmployees generate petabytes of data about them selves everyday, says Waber. But that data sits in disparate systems in dif-ferent formats an d is often messy. To make it work, you needaccess to all of this information in real time, Waber says. ITis the backbone for this en tire process.

Implementing a single version of an HR information systemitself may not sound revolutionary, but it's a critical first step forcompanies interested in more advanced analytics.

Jo Stoner, senior vice president of worldwide HR for Infor-

mtica, knew predictive talent analytics could benefit thegrowing data-integration company. A lot of companies don'tmake it past a billion [in revenue]. We were starting to hit thoseawkward teenage years, she explains. Managing the compa-

SEPTEMBER 1 2013 j ww w.cio .com

8/12/2019 Invasion of the Data Scientists.

5/8

COVER STORY :: n a l y t i c s

ny's assets would be critical to maintaining mom entum. But we don't own buildings or raw materials, says Stoner. Ourgreatest asset is our talent. First, though, the company had to 'bring all its HR data together, applying the master data manag e-

ment services Informtica delivers to clients to its own intern alemployee data in ord er to layer analytics atop it.

For most companies, just arriving at a single version of theHR tru th can be beneficial. Paul Lones, senior vice president ofIT at Fairchild Semiconductor, says that two years ago, manag-ers at the chip m aker lacked a single system tha t could providean accurate tally of employees worldwide, let alone show theamo unt of employee turnover. Reports had to be com piled frommu ltiple systems. Succession plann ing took place in MicrosoftWord docum ents. Compensation decisions might be made inisolation. Now that the company ha s implemented cloud-basedWorkday, managers can access data on all 9 000 employees

in one place, including succession plans, turnover trends andsalary information. A m anager in the Philippines consideringa raise and promotion for an employee can see in seconds howthat will compare with others in the group and w ith local com-pensation trends and m ake that decision, says Lones.

It may not be rocket science, but it's a startone that's beena long time coming for m any HR groups. Chiquita Brands, forexample, had multiple homegrown and manual HR systems.

It was a cobbled-together thing, says Kevin Ledford, Chiq-uita's CIO. People spent 90 percent of their time figuring outwhere the data was and 10 percent on analyzing it. In 2008, thecompany m oved to a global HR system, which came in han dywhe n Chiquita moved its head quarters from Cincinnati, Ohio,

H Lto deliver Dusiness results.And the language of th ebusiness is ana iytics,

-Michae l Arena , D i rec tor of lobal Ta len t and

Orga nizationa l Capability, General Motors

buildings, to staff c ent projects m ore efficiently and effectively. In the past we couldn't tell who was mobile, says Endry. Nowwhen we have a giant project in Ohio, we can see on a dashboardthat we've got these three people in Boston willing to move there.

Marc Franciosa, CIO of Praxair, has tied the company's HRand employee performance systems to non-HR systems likeSharePoint as a foundation for the company's talent analyt-ics initiativeno small task for the $11 billion indus trial andmedical gases company with 26 ,000 employees in 50 coun-tries. The underlying data and p rocesses have to be consis-tent to be able to do any real analytics with confidence, saysFranciosa. For companies that are fairly mature that haven'thad a global environment before, it's going through that initialnormalization and standardization process to make sure thatthis certification, for example, means the sam e thing aroundthe world, says Franciosa. (He implemented SumTotal's HRmanagement system and ElixHR platform to link disparatedata.) The cleanup has been a challenge.

Now, when Praxair w ants to make a bid or sign a new cus-tomer, managers can analyze HR implications first. Do theyhave people who speak Portugue se, have the necessary certi-fication , and are w illing to relocate to Rio de Janeiro? We cando some m odeling of the skill sets to determ ine if it's doable orif we will have to recruit externally, Franciosa says.

A t GM, A rena has been implementing a three-phase analyt-ics plan. First, integrate systems in a way tha t ens ures highlyaccurate data is available. Next, push much of that d ata intostandardized repo rting tools and dashboa rds that bu sinessmanagers can use on their own. Then start building models.

One of the first projects A rena imple-mented was a means-based comparisonanalys is of the top talent pool. The modelexamines every employee data field n thePeopleSoft databa se to look for im portan tinsights. A rena says. Five or six experi-ences may jum p o ut Having internationalexperience ma y statistically matter. Thenwe dig deeper. A re there certain types ofinternational experiences that mattermore than others? Does that need to hap-pen earlier versu s later?

fo Charlotte, N.C., and lost 75 percent of its corpora te employ-ees. It was very tumultuous. We threw all of our monkeysin the air, and the y all came down in different buc kets, saysLedford. It would have been a nightmare [without the globalHR system]. Now that the company is exploring predictiveHR analytics, that success with master data management iseverything, says Ledford.

A t A rcadis, Endry has connected his cloud-based workforce-

management system to 11 other pieces of software, includingERP, learning managem ent, payroll and an active directory. Thecombined data helps the company, which provides engineer-ing services regarding infrastructure, water, environment and

Divining Interven t ion sThe real power is in applying predic-tive analytics to a corporate population.

Everyone's talking about it, says Chiquita's Ledford, lookingat all this data you have and trying to figure out the future.

The typical data warehouse approach is looking back, butwhat we wanted to do wa s start looking forward, says Praxair'sFranciosa. What are the leading indicators we should be look-ing for? What are those m etrics or data sets we d on't have but, ifwe did, would really help us? Wha t external data sources couldwe us e to drive better decision-making?

For example, Praxair is growing by double digits in China. Rather than hiring a ton of people and trying to recreate thewheel [there], what I've been driving is how do we replicaterapidly those things that have made us successful in our

SEPTEMBER 1, 2013 www .cio .co m

8/12/2019 Invasion of the Data Scientists.

6/8

mature geographies, says Franciosa. There's a huge opportunity to usepredictive analytics based on wherewe're best-in-class.

The predictive analytics marketfor HR is nascent and wide-open. We partner w ith them all, from IBMto SuccessFactors to PeopleSoft,says GM's Arena. They're all tryingto play in the space, bu t I don't knowthat any of them have figured t out.

Arena's team has built a modelthat predicts what changes inattrition rates will mean for GM'sworkforce. Previously, if someoneproposed hiring a bun ch of young engineers, no one could becertain if that w as the best decision. Now we can say, let's seewhat that looks like five years from now, Arena says. Whatare the dividends if we hire 2 00 entry-level engineers? Mightwe be better off h iring SO advanced engineers? We can takethat information to the head of engineering and say, 'Here'swh at it will cost you.'

Arena thinks that analyzing the interactions of networksof employees holds the m ost promise. The process starts witha survey. We ask questions of a given network: Who do yougo to when you want to shop a new idea? Where do you turnwhen you need resources to get things done? Then we run theanalytics, Arena explains. We can tell you who the brokersare, who's central in that network, w ho are the bridges acrosssilos. We can even predict wh o's a flight risk based on w herethey sit in the network. And by identifying which employeenetworks are most productive. Arena says there's a chance toimprove performance across the company.

At Coca-Cola Enterpr ises, Crum ley is integrating bus inessdata with HR data for predictive purpo ses. That's where youcan really get sexy with it, he says. WMle working with IT toclean and standardize all the data, Crumley is partnering w itheach corporate function to find out what business metric mightbe the key meas ure of success for their em ployees. By combin-ing those business m etrics with people data, he hopes to be ableto reverse engineer what a successful employee is, so we can

get the best candidates in the future.Employee engagement is a leading indicator of talent reten-

tion at Coca-Cola Enterprises. And one of the biggest boosters ofemployee engagement num bers is access to on-the-job learning,so Crumley's team is trying to figure out how to make train-ing oppo rtunities more un iversal. For example, why are folksin this shift at this plant not taking classes as much as o theremployees in that line of business? W ith answ ers to questionslike that, HR can intervene to address the core reason, whetherthat's an accessibility problem or a manager w ho needs m orecoaching. Crumley says the effort w ill gain even mo re steamwhen HR is able to show, throu gh data analytics, a co rrelationbetween taking a specific training course and an improvementin sales or productivity.

At call-center provide r NOVO 1, CTO Mitchell Swindell hasimplemented a predictive hiring tool from Evolv. Applicants

-David Crumley Vice President of lobal Human Resources Coca-Cola Enterprises

complete a Web-based application that screens for attitude,

propensity for customer service, and voice capabilities. Thesoftware also shows the candidate what it's like to work in a callcenter in hopes of screening out those who would be a poor fitin the high-turnover indu stry The tool then gives the candidatea red, yellow or green rating, at which point cand idates ratedgreen or yellow are invited for in-person interviews. The hiringdecision is still in the han ds of a hum an, but the system has pre-dicted with 8 0 percent accuracy the company's top performers,based on 90-day follow-up data on the hired employees. Sinceintroducing the algorithm-enhanced h iring system, tenure is upby 2S percent, agent productivity has increased 30 percent, andthe overall staffing bu dget h as decreased 11 percent. Swindellhas integrated Evolve with the company's payroll, workforce-

management and p roprietary quality systems to help developa more nuanced profile of the best employees.

At Chiquita, Ledford is exploring predictive analytics to helpthe company find, train and retain its bananaeros expertsin growing bananas. Those guys a re really hard to find, asbananas have taken a backseat to coffee and tourism, says Led-ford. Analytics could enable manag ers to predict which lower-level employees could become our next wave of bana na folks,says Ledford, and determine the right training and groomingto make that happen.

Employee TrackingThere's also a gold mine of information in how people movethrough an organization, and a handful of comp anies are look-ing at p hysically track ing employeesoften via RFID -enabledbadgesto find out how people work and what impact that canhave on business outcomes.

The barrier at this point is not the technology, says Waber,whose Sociometric Solutions is an early provider of sensor-based analysis. I can tell you how much more money a com-pany makes when two employees eat lunch together. We can doextremely sophisticated things. The challenge is that organiza-tions are not used to looking at themselves th is way.

When GM's Arena was senior vice presiden t of leadership

development at Bank of America in 2010, the financial ser-vices com pany used sensors to track 90 call-center workersover the course of several weeks and found that those in themost cohesive network s were the m ost productive. By switching

www.cfo .com i SEPTEMBER 1 20 13

8/12/2019 Invasion of the Data Scientists.

7/8

COVER STORY :: n a l y t i c s

from solo to group b reak times, enco uraging mo re socialization,agents improved efficiency by 10 percent. As silly as it sounds,it worked, says Aren a. The analytics told us it was prob ablythe right thing to do. Sometimes it's as simple as moving desks

closer together, says Waber Steelcase's SuUivan has discoveredthat the size of lunch tables can have an impact on productivity.You can't force peo ple to interact m ore, says Waber, but bas edon the data, you can engineer serendipity.

Although A rena conducted a num ber of experiments usingsensor data at BofA, he's not quite ready to start tracking work-ers at GM. I'm a huge advocate of sensor work, Are na says.

But it's laden with trust and privacy issues and a lot of organiza-tions just aren't ready for that. It can be a b it of a slippery slope.

Praxair is conducting a pilot using sensors on its remoteworkers. The system will measure how long it takes a workerto, say, install a tank for a customer, by m onitoring their move-

ments via a sensor on their protective equipment. The sensoralso monitors wo rkers for exposure to harmful gases. If gasis detected, an alarm goes off and the monitoring center willattempt to comm unicate w ith the worker. Franciosa envisionsintegrating the sensor data into other corporate systems touncover correlations between events and particular locations,types of employees, or certifications.

The Importance of TransparencyFranciosa expects employees to put up some resistance tobeing physically tracked, much like the pushback the companyencountered w hen it was first placing computers onboard itstrucks . It was viewed as Big Brother wantin g to know how

warehousp

fast I drive or how har d I brake, says Franciosa. The way toalleviate that is transparency. People w on't Hke being physicallymonitored if they think w e're trying to find out how long theirbreak was. So we have to be completely transp arent that weare using this for safety and long-term p roductivity. They'llrecognize the value in that.

HR collects all kind s of sensitive employee information, bu temployees see physical tracking as particularly intrusive. It ist

bo un dar y to cross, says Steelcase's Sullivan. All of Steel-case's sensor-related experiments are opt-in. Company analystssee only aggregate data, not individual histories. An d Sullivan'steam com municates the process and the intentions not just to

those who have signed up , but also to everyone on the cam pus. In the U.S., employees don't really legally have pro tections

around this data. A company can track you w herever you goand listen to all your conversations, says Waber. But that

defeats the purpo se of this approach, which is trying to helppeople work better, be happier an d stay at their jobs.

Comm unication is critical with any collection and analysisof people datanot just sensor d ata. I don't think we're doinganything that people haven't been trying to do for years, saysInformatica's Stoner. But we have to say wh at we will do withthat data.

Praxair's Franciosa has a close partn ership with h is legalteams aroun d the wo rld to navigate the various data privacyand protection issues in each country. But even once we unde r-stand that we can have this data, we have to be very transpa rentand say, here's why we wan t your picture or your talent profile,Franciosa says. That goes

a long way toward gaining both cred-

ibility and traction.

The Role of Data in the People Business What's really happening right now is a shift in HR from an artto a science, says Crum ley of Coca-Cola Enter pris es, who 'scurre ntly exploring how social netwo rk data and gamificationmight become part of his HR analytics platform. A lot of HRteams are trying to figure out how to mak e that shift qu ickly soit's no longer HR sitting aroun d w aiting to be pulled in, but H Rcoming to the table w ith nu ggets of wisdom.

Data analytics could enable HR to elevate itself from a tacti-cal suppo rt function to a business p artner on strategy, which

ought to sound p retty familiar to CIOs.But there are limits to H R's data-driven

transforma tion. [Analytics] are all aboutprobability, and there's just so far you cango with probability, says Crumley. If youwant to figure out how m any employeesyou need to launch a new product, it canget you in the right ballpark. When itcomes to predicting turnover, it's not anexact science. People are people.

It 's never black-and-white whenyou're talking about people, says Stoner

of Informtica. W hue some folks get star sin their eyes when talking about big data,Stoner often sees a bigger haystack to siftthrough. But analytics, she says, help

point companies in the right direction. In HR, we live in aworld w here data b rings m ore questions. You always have tolook beneath it, she says. It's not an exact science. But at leastit gets us looking at the right p art of the haystack so we can getto the answe r faster.

That's why GM's Arena says his talent analytics will neverbe fully auto mated . Sometimes we get projections wro ng forall kinds of reason s. It can take several iterations. But HR still

loves it, because it equips them to m ake intelligent decisions fortheir business partners. QEI

Stephanie Overby is freelance writer based in Massachusetts.

SEPTEMBER 1 2013 ww w.cio.c om

8/12/2019 Invasion of the Data Scientists.

8/8

C o p y r i g h t o f C I O i s t h e p r o p e r t y o f C X O M e d i a I n c . a n d i te m a i l e d t o m u l t i p l e s i t e s o r p o s t e d t o a l i s t s e r v w i t h o u t t h e p e r m i s s i o n . H o w e v e r , u s e r s m a y p r i n t , d o w n l o a d , o r e m a i