k

by Thomas H. Davenport;, Jeanne Harris, and Jeremy Shapiro

040 HBR.ORG

Do you think you knowhow to get the bestfrom your people? Ordo you know? How doinvestments in youremployees actuallyaffect workforceperformance? Who areyour top performers?How can you empowerand motivate otheremployees to excel?

October 2010 Harvard Business Review 5 3

COMPETING ON TALENT ANALYTICS

Applying Talent Analytics

Leading-edge companies are increasingly adopt-ing sophisticated methods of analyzing employee

data to enhance their competitive advantage . Google,Best Buy, Sysco, and others are beginning to under-stand exactly how to ensure the highest productivity,engagement, and retention of top talent, and thenreplicating their successes . If you want better perfor-mance from your top employees-who are perhapsyour greatest asset and your largest expense-you'lldo well to favor analytics over your gut instincts .

Harrah's Entertainment is well-known for em-ploying analytics to select customers with the great-est profit potential and to refine pricing and promo-tions for targeted segments . (See "Competing onAnalytics," HBR January 2006 .) Harrah's has also ex-tended this approach to people decisions, using in-sights derived from data to put the right employeesin the right jobs and creating models that calculatethe optimal number of staff members to deal withcustomers at the front desk and other service points .Today the company uses analytics to hold itself ac-countable for the things that matter most to its staff,knowing that happier and healthier employees cre-ate better-satisfied guests.

For example, Harrah's used metrics to evalu-ate the effects of its health and wellness programson employee engagement and the bottom line .Preventive-care visits to its on-site clinics have in-creased, lowering urgent-care costs by millions ofdollars over the past 12months. And because Harrah'sunderstands the relationship between employee en-gagement and top-line revenue, it can evaluate theprogram according to revenue contribution as well .

Here's how other organizations use analytics toimprove their management of human capital :

• Almost every company we've studied says itvalues employee engagement, but some-includingStarbucks, Limited Brands, and Best Buy-can pre-cisely identify the value of a o.1%increase in engage-ment among employees at a particular store . At BestBuy, for example, that value is more than $100,000

in the store's annual operating income .• Many companies favor job candidates with stel-

lar academic records from prestigious schools-butAT&T and Google have established through quantita-tive analysis that a demonstrated ability to take ini-tiative is a far better predictor of high performanceon the job .

• Employee attrition can be less of a problem whenmanagers see it coming. Sprint has identified the fac-tors that best foretell which employees will leave af-

54 Harvard Business Review October 2010

Six kinds of analytics canhelp companies answercritical talent questions-from the simple ones atthe bottom of this ladderto the more sophisticatedones at the top .

WHY DO EMPLOYEESCHOOSE TO STAYWITH-OR LEAVE-MY COMPANY?Google suspected thatmany of its low-performingemployees were either mis-placed in the organizationor poorly managed . Em-ployee performance databore that out .

WHICH ACTIONS HAVETHE GREATEST IMPACTON MY BUSINESS?By keeping track of thesatisfaction levels of deliv-ery associates, Sysco im-proved their retention ratefrom 65% to 85%, savingnearly $5o million in hiringand training costs .

WHAT ARE THE KEYINDICATORS OF MYORGANIZATION'SOVERALL HEALTH?JetBlue analystsdeveloped a metric-the "crewmember netpromoter score"-thatmonitors employee en-gagement and predictsfinancial performance .

TALENTSUPPLY CHAIN

∎TALENTVALUE MODEL

∎

WORKFORCEFORECASTS

∎HUMAN-CAPITALINVESTMENTANALYSIS

∎

ANALYTICAL HR

HUMAN-CAPITALFACTS

HOW SHOULD MYWORKFORCE NEEDSADAPT TO CHANGESIIV THE BUSINESSENVIRONMENT?Retail companies canuse analytics to predictincoming call-centervolume and releasehourly employees earlyif it's expected to drop .

HOW DO I KNOWWHIEN TO STAFF UPOR CUT BACK?Dow Chemical has acustom modeling tool thatpredicts future head countfor each business unit andcan adjust its predictionsfor industry trends, politi-cal or legal developments,and various "what if"scenarios .

WHICH UNITS,DEPARTMENTS,OR IINDIVIDUALSNEED ATTENTION?Managers at LockheedMartin use an automatedsystem to collect timelyperformance-review dataand identify areas needingimprovement .

Idea in Briefl-eaaing companiessuch as Google, BestBuy, P&G, and Sysco usesophisticated data-collection technologyand analysis to get themost value from theirtalent .

These companies have takenthe guesswork out of employeemanagement by leveraginganalytics to improve theirmethods of attracting and re-taining talent, connecting theiremployee data to businessperformance, differentiatingthemselves from competitors,and more.

ter a relatively short time . (Hint : Don't expect a longtenure from someone who hasn't signed up for theretirement program .)

• Professional sports teams, with their outsizeexpenditures on talent, have been leading users ofanalytics . To protect its investments, the soccer teamAC Milan created its own biomedical research unit .Drawing on some 60,000 data points for each player,the unit helps the team gauge players' health and fit-ness and make contract decisions .

What's driving this shift to analytics? Certainly,companies today want more from their talent . That'swhy some are reinventing a whole range of peoplepractices : Netflix has tossed aside traditional HRabsence policies, and Best Buy's corporate office es-chews standard work schedules. Analytics takes theguesswork out of fresh management approaches . Atthe same time, voluminous "digital trails" of datafrom knowledge management systems and socialnetworks are now available for analysis . The publicrelations firm Ketchum, for example, analyzed per-sonal networks in its London office to learn how eas-ily information flowed across teams . Cognizant, aU.S.-based professional services firm with many em-ployees in India, analyzed social media contributions,particularly blogs . It found that bloggers were moreengaged and satisfied than others and performedabout 10% better, on average .

At Best Buy the value of a o.1%increase in employee engagementat a particular store is $ioo,ooo .

In this article are six key waysto track, analyze, and useemployee data : They rangefrom establishing simplemetrics that monitor yourorganization's overall healthto identifying talent shortagesand excesses long before theyhappen .

In our work with companies like these, we haveseen best practices emerge for using analytics tomanage people .

Six Uses of Talent AnalyticsAnalyzing talent is not significantly different fromanalyzing customer relationships or supply chainmanagement. It starts with the delivery of historicalfacts ("What happened?") and ends with real-timedeployment of talent based on rapidly changingneeds. The six kinds of analytics for managing yourworkforce, from simplest to most sophisticated, arehuman-capital facts, analytical HR, human-capitalinvestment analysis, workforce forecasts, the talentvalue model, and the talent supply chain .

Human-capital facts are a single version ofthe truth regarding individual performance andenterprise-level data such as head count, contin-gent labor use, turnover, and recruiting. Companiesshould carefully consider what facts will give themthat version . For some, one or two data points mayindicate overall health . For example, JetBlue cre-ated an employee-satisfaction metric around itspeople's willingness to recommend the companyas a place to work. This "crewmember net promoterscore" (modeled after the customer-satisfactionmetric) has been used to study the impact of com-pensation changes and to help determine executive

HBR.ORG

Companies that want to com-pete on talent analytics musthave access to high-qualitydata and manage them at anenterprise level, support ana-lytical leaders, choose realistictargets for analysis, and hireanalysts with a broad base ofexpertise .

October 2010 Harvard Business Review 55

COMPETING ON TALENT ANALYTICS

Google's highly analytical culture and practicesextend to its human resources function .

The company's goal is toidentify leading people-management practices andconfirm them with data andanalysis . To achieve it, Googlecreated a people analyticsfunction with its own directorand a staff of 3o researchers,analysts, and consultants who

HBR.ORG Do you havequestions or commentsabout this article? TomDavenport will respond toreader feedback at hbr.orgthrough mid-October.

56 Harvard Business Review October 2010

study employee-related deci-sions and issues . The Peopleand Innovation Lab (PiLab)conducts focused investiga-tions for internal clients.

Google has analyzed a vari-ety of HR topics and has oftenmoved in new directions as aresult. It has determined what

bonuses. Employees are asked annually on their hir-ing date if they would recommend the company, soJetBlue can effectively monitor employee engage-ment monthly.

JetBlue and other successful organizations aretransparent with end users about the process: Anymanager or employee may see how the data werecollected, what formulas are being used, and, mostimportant, why the data matter to the operation . Forexample, Harrah's provides documentation in its HRscorecard to ensure that all readers understand howhuman-capital facts are created and what they meanfor daily management .

Analytical HR collects or segments HR data togain insights into specific departments or functions .For example, a manager might be able to see thatstaff-turnover intervention is needed for the EastCoast sales team but not the West Coast team. Analyt-ical HR integrates individual performance data, suchas personal achievement in key result areas, with HRprocess metrics, such as cost and time, and outcomemetrics, such as engagement and retention .

Lockheed Martin built a performance manage-ment system to link each employee's performanceto organizational objectives . The automated systemcollects timely performance-review data throughoutthe year. The data can then be compared with knowl-edge management information, such as who hasundergone formal training in specific areas . With thesystem, Lockheed Martin can identify its high poten-tials for special programs or monitor employees whoneed improvement in certain areas .

Human-capital investment analysis helpsan organization understand which actions have thegreatest impact on business performance . One leader

in this area is Sysco, the $36 .8 billion Fortune 100global food-service company. Sysco is a complex or-ganization made up of nearly a hundred autonomousoperating units and about 51,000 full-time employ-

backgrounds and capabili-ties are associated with highperformance and what factorsare likely to lead to attrition-such as an employee's feelingunderused at the company .It has set the ideal number ofrecruiting interviews at five,down from a previous averageof ten .

Google's Project Oxygen-sonamed because good manage-

ees serving approximately 400,000 customers . Thecompany began its workforce analysis with threegross measures for each operating unit: work climateand employee satisfaction, productivity, and reten-tion. It has drilled deeper to understand, measure,and manage seven other dimensions of the work en-vironment, including frontline supervisor effective-ness, diversity, and quality of life .

Sysco's analysis revealed that operating units withhighly satisfied employees have higher revenues,lower costs, greater employee retention, and supe-rior customer loyalty. The company can efficientlyidentify what actions by management will have thegreatest impact on the business . For example, in sixyears it has improved the retention rate for deliveryassociates-who provide customer service and buildcustomer relationships-from 65% to 85%. Syscotracks the group's satisfaction scores, and whenthey dip, it institutes immediate improvements toget them back on track. By retaining this key talent,Sysco saved nearly $50 million in hiring and trainingcosts for new associates .

Workforce forecasts analyze turnover, succes-sion planning, and business opportunity data to iden-tify potential shortages or excesses of key capabilitieslong before they happen . As Vinay Couto, Frank Ri-beiro, and Andrew Tipping wrote recently in Strategy+ Business, Dow Chemical has evolved its workforceplanning over the past decade, mining historical dataon its 40,000 employees to anticipate workforceneeds throughout the chemical industry's volatilebusiness cycles. It forecasts promotion rates, internaltransfers, and overall labor availability . Dow uses acustom modeling tool to segment the workforce intofive age groups and 10 job levels and calculates futurehead count by segment and level for each businessunit. These detailed predictions are aggregated toyield a workforce projection for the entire company .Dow can engage in "what if" scenario planning, al-

ment keeps the companyalive-was established todetermine the attributes ofsuccessful managers. ThePiLab team analyzed annualemployee surveys, perfor-mance management scores,and other data to dividemanagers into four groups ac-cording to their quality. It theninterviewed high- and low-scoring managers (interviews

were double-blind-neither in-terviewers nor managers knewwhich category the managerswere in) to determine theirmanagerial practices . Googlewas eventually able to identifyeight behaviors that character-ized good managers and fivebehaviors that all managersshould avoid .

Google's vice president ofpeople operations, Laszlo

tering assumptions on internal variables such as staffpromotions or external variables such as political andlegal considerations. Workforce forecasts can be usedto staff up in key growth areas or identify knowledgemanagement risks for retiring employees before theyare clear to managers .

The talent value model addresses questionslike "Why do employees choose to stay with ourcompany?" A company can use analytics to calcu-late what employees value most and then create amodel that will boost retention rates . Such a modelcan help managers design personalized performanceincentives, assess whether to match a competitor'srecruitment offer, or decide when to promote some-one. Google uses employee performance data to de-termine the most appropriate intervention to helpboth high- and low-performing employees succeed .Laszlo Bock, Google's vice president of people opera-tions, told us, "We don't use performance data to lookat the averages but to monitor the highest and low-est performers on the distribution curve . The lowest5% of performers we actively try to help . We knowwe've hired talented people, and we genuinely wantthem to succeed ." The company's hypothesis wasthat many of these individuals might be misplaced orpoorly managed, and a detailed analysis supportedthat idea. Understanding individuals' needs and val-ues allowed Bock's team to successfully address anumber of difficult situations .

The talent supply chain helps companies makedecisions in real time about talent-related demands-from optimizing a retail store's next-day work sched-ules, on the basis of predicted receipts and indi-viduals' sales performance patterns, to forecastinginbound call-center volume and allowing hourly staffmembers to leave early if it's expected to drop . Thisis the most complex of the six kinds of talent analyt-ics, because it requires particularly high-quality data,rigorous analysis, and the integration of broad talent

Bock, says, "It's not thecompany-provided lunch thatkeeps people here . Googlerstell us that there are threereasons they stay : the mis-sion, the quality of the people,and the chance to build theskill set of a better leader orentrepreneur. And all our ana-lytics are built around thesereasons ."

management and other organizational processes .Talent supply chains are still in their infancy, but theearly success of some organizations, particularly inthe retail space, suggest that they will spread .

Mastering Talent AnalyticsUnsurprisingly, building a capability in this domainrequires the same fundamentals that most otherbusiness analysis does. We summarize them withthe acronym Delta (access to high-quality data, en-terprise orientation, analytical leadership, strategictargets, and analysts) .

Data . Organizations can get increasingly good HRdata from their enterprise systems, but they some-

Cognizant's analytics revealedthat employees who blogged weremore engaged and satisfied .

times need to augment them with new metrics, likeJetBlue's . At Harrah's many line managers, who arealready on the floor at its properties, observe and re-cord the frequency with which customer-facing staffmembers smile, because that behavior is highly cor-related with customer satisfaction . Data needn't beperfect to be appropriate for analysis-just sufficientto understand trends that matter .

Enterprise . HR can no longer confine employeedata to its silo ; organizations need access to thosedata to be successful . JetBlue, Best Buy, and LimitedBrands have observed an important statistical rela-tionship between employee satisfaction and com-pany performance-usually at the station, branch, orstore level . The significance of the relationship mo-tivated Best Buy to make its employee engagementsurveys quarterly rather than annual .

H BR.ORG

October 2010 Harvard Business Review 57

COMPETING ON TALENT ANALYTICS

Common Mistakes in Talent Anal icsCompanies that use analytics for employeemanagement can create tangible value forthemselves as long as they avoid these mistakes :

~lr Making analyticsan excuse to treathuman beings likeinterchangeablewidgets

~k Keeping a metriclive even when it hasno clear businessreason for being

~Ilr Relying on just a fewmetrics to evaluateemployee perfor-mance, so smartemployees can gamethe system

Dow minesemployee datato forecastpromotion ratesand internaltransfers .

58 Harvard Business Review October 2010

~k Insisting on loo%accurate databefore an analysisis accepted-whichamounts to nevermaking a decision

~k Assessing employeesonly on simple mea-sures such as gradesand test scores,which often fail toaccurately predictsuccess

Sir Using analytics tohire lower-levelpeople but not whenassessing seniormanagement

91r Failing to monitorchanges in organiza-tional priorities, thuscreating irrelevant-if accurate-analyses

~k Ignoring aspects ofperformance thatcan't easily be trans-lated into quantita-tive measures

~Ir Analyzing HR ef-ficiency metricsonly, while failing toaddress the impactof talent manage-ment on businessperformance

Leadership . The success of almost any initiativedepends on its leaders, and talent analytics is no ex-ception. In fact, at the organizations we've researchedand worked with, leaders' commitment to this ap-proach is the single most important factor in whetherit succeeds . Because the data pertain to human behav-ior, executives maybe skeptical . Comcast's senior vicepresident of compensation and benefits, Bill Strahan,recalls, "It was crucial for manager adoption that wepresent the analytics business case in the language ofour company, focusing on competitive pressures andthe people component of our change"

Leaders who believe that human-capital insightsshould be used to solve business problems mustconstantly press for decisions and analyses based onfacts and data rather than on tradition, hearsay, orsupposition . And they should foster a culture thatallows for experimentation and mistakes-which areoften unacceptable in HR functions today .

Targets . Organizations that use talent analyticshave already made people the focus of analytical

activity. But should they concentrate on hiring, as-signments to projects and tasks, or retention? Whichtypes of employees need the most analytical atten-tion? Which of the six kinds of talent analytics shouldbe employed when? When Google was adding looemployees a week, from 2005 into 2008, hiring the

right people was its primary focus . When hiring

HBR.ORG

slowed in 2008 and 2009, the company turned togaining insights into employee attrition and effectivemanagement approaches.

Analysts . Analytical theory must be convertedinto practice . This requires experts not only in quan-titative analysis but also in psychometrics, humanresource management systems and processes, andemployment law. Industrial-organizational psy-chologists are especially helpful in creating analyti-cal initiatives and ongoing programs . Google, P&G,Royal Bank of Scotland, Intel, and Tesco have all es-tablished HR analytics groups to get deeper insightsinto their people practices.

The best analysts can persuade managers to adoptanalytical decision making . In late 2009 Harrah's be-gan recruiting an external sales force and used orga-nizational psychologists to create a predictive assess-ment for the job . But during the interview processmanagers became emotionally attached to some ofthe candidates with low probabilities of success . Theanalysts were prepared: They used randomized test-ing to prove that analytics was the superior method,and relied on their interpersonal skills to sway deci-sions when necessary. One management team at atroubled Harrah's location was astounded by thehigh call volume and conversion rates the new hiresachieved, which helped reverse a decline in sales .

No organization we've worked with has embracedan analytics-only method of managing, motivating,and retaining employees. But early adopters havecreated tangible value for themselves by applyingthe right data and tools to people processes . The bestorganizations see their people not only as individualsbut also as a rich source of collective data that manag-ers can use to make better decisions about talent .

Future organizational performance is inextricablylinked to the capabilities and motivations of a com-pany's people. Organizations that have used data togain human-capital insights already have a hard-to-replicate competitive advantage. Others, too, candraw on these new techniques to improve their busi-ness results . ~7

HBR Reprint R1010B

Thomas H . Davenport ( tdavenport@babson .ed u ) isthe President's Distinguished Professor of informa-

tion Technology and Management at Babson College andthe author, coauthor, or editor of 13 books. Jeanne Harris(j eanne .g.harris@accenture.com) is an executive researchfellow and a senior executive at Accenture's Institute forHigh Performance . She and Davenport are the coauthors ofAnalytics at Work (Harvard Business Review Press, 2010) .Jeremy Shapiro (j eremy.shapiro@morganstantey.com) isan executive director in human resources at Morgan Stanleyand a coauthor of Ultimate Performance (Wiley, 2007) .