Diamonds

BEST PRACTICE

Diamondsin the

DataMineby Gary Loveman

Harrah's Entertainment has outplayed its

competition and won impressive gains, despite

being dealt a weak hand by the economy

The secret? Mining the company's rich database

to develop compelling customer incentives.

I T'S A FRIDAY NIGHT on the Las VegasStrip, and all of the neighbors aremaking spectacles of themselves. The

$750 million Mirage boasts a Vesuvianvolcano that erupts loudly every 15 min-utes. Next door, at Treasure Island, afaux British frigate battles a pirate shipat regular intervals. Further down theStrip, the Bellagio sports a lake fes-tooned with sparkling, dancing foun-tains that beckon to passing tourists.

Meanwhile, the customer pulling IntoHarrah's Las Vegas is dazzled more bythe service than the building. A smilingvalet greets her by name. Instead ofhaving to wade through a crowdedlobby to reach the casino, she stepsquickly into the gaming room and sitsdown at a slot machine. The card readeron the machine pages her host, who ap-proaches every so often to ensure that

MAY 2003

she's happy with the service she's re-ceiving. Although the customer doesn'tfit the stereotypical profile of a Las Vegashigh roller, Harrah's makes sure shefeels special. Because the casino deliversthe recognition and service she hascome to expect, she'll return to Harrah'sagain and again.

Harrah's Entertainment has the mostdevoted clientele in the casino indus-try-a business notorious for fickle cus-tomers. That loyalty to Harrah's, whichnow operates 26 casinos in 13 states, hasenabled the gaming company to recordT6 straight quarters of same-store rev-enue growth. In 2002, Harrah's postedmore than $4 billion in revenue and$235 million in net income.

We've increased customer loyalty,even in the current challenging econ-omy, in two ways. First, we use database

109

BEST PRACTICE • Diamonds in the Data Mine

marketing and decision-science-basedanalytical tools to widen the gap be-tween us and casino operators who basetheir customer incentives more on intu-ition than evidence. Second, we deliverthe great service that consumers de-mand. In short, we've come out on topin the casino wars by mining our cus-tomer data deeply, running marketingexperiments, and using the results to de-velop and implement finely tuned mar-keting and service-delivery strategiesthat keep our customers coming back.

A Dicey BusinessBy the time I had left Harvard BusinessSchool to Join the corporation as chiefoperating officer in 1998, Harrah's hadbecome the first nationwide casino busi-ness, thanks to a geographic diversifi-cation plan promulgated by Phil Satre,the company's chairman and then-chiefexecutive officer. Satre led the wave ofgaming growth in the 1990s, expandingHarrah's from four casinos in two statesto 26 casinos across 13 states currently.

Satre's vision differed markedly fromthe strategy pursued by other big casinooperators, whose "If you build it, theywill come" philosophy focused on at-tracting customers to a fantasyland LasVegas. Companies such as MandalayResort Group and MGM-Mirage in-vested heavily in constructing costlymust-see casinos offering a wide rangeof amenities-fabulous spas, high-endshopping malls, dazzling shows-de-signed to appeal to a broader audiencethan simply gamblers. Their hope wasthat such facilities would attract an ever-growing number of new customers. Thisstrategy ultimately transformed boththe Las Vegas skyline and tourist spend-ing patterns: The Las Vegas Conventionand Visitors Authority reported that, in2001, revenues from dining, entertain-ment, shopping, and other activitiesoutpaced the city's casino revenues bya three-to-one margin.

Satre, by contrast, focused on ex-panding the corporation's gaming busi-

ness outside Nevada and Atlantic City,seeing geographic diversification as anopportunity to introduce the Harrah'sbrand to new customers and to insulatethe company from regional economicvagaries. Once Harrah's posted high re-turns in these emerging markets, itscompetitors also began to expand intothem. But Satre believed that compet-ing largely on the basis of hillion-doUarfacilities in the face of new competitionwas not the most prudent use of capitalbecause the returns on such buildingsoften weaken when the novelty wanes.

Fortunately, Satre had two importantarrows in his quiver. First, he knew that,unlike its competitors, Harrah's didn't

Former Harvard Business School professor Gary Loveman is the CEO of Harrah'sEntertainment. He is the author or coauthor of five HBR articles, Including "Puttingthe Service-Profit Chain to Work"(March-April 1994)-

depend heavily on its stores, restau-rants, bars, or shows; it drew the lion'sshare of its revenues-87.2% in 2001-from its casinos. He also suspected thatcultivating lasting relationships withthe company's core customers-slotplayers - would lead to greater and moresustainable profit growth. So he optedto invest in development of the intel-lectual and technological capabilitiesneeded to assemble and analyze dataabout those customers. The goal was toprovide good service to them and thusencourage their loyalty to the company'sbrand. When Satre hired me as COO, hesaid he wanted to change Harrah's froman operations-driven company thatviewed each casino as a stand-alonebusiness into a marketing-driven com-pany that built customer loyalty to allHarrah's properties.

l ip HARVARD BUSINESS REVIEW

Diamonds in the Data Mine • BEST PRACTICE

One tactic the company had alreadydecided to use to enhance customerloyalty was called Total Gold, a player-card program that was modeled afterthe airline industry's frequent-fiier pro-grams. Launched in 1997, Total Gold wasdesigned to provide regular customerswith incentives to visit Harrah's prop-erties throughout the country. Custom-ers inserted their Total Gold cards intoslot machines and earned credits asthey played. They were rewarded withthe standard fare that all casinos offer-free hotel rooms, dinners, show tickets,gift certificates. But there were threeproblems with the program. First, noth-ing differentiated this program fromour competitors' efforts. Our customerssimply took their free rooms and din-ners and drifted across the street to dotheir gambling. Second, our customersearned different rewards at differentproperties; there was no uniformity inthe program. Third, and most impor-tant, our customers were not given anyincentives to consolidate their gamingwith Harrah's.

While Total Gold wasn't much goodfor keeping customers loyal to Harrah's,it was quietly digging our future dia-mond mine. By tracking millions of in-dividual transactions, the information-technology systems that underlie theprogram had assembled a vast amountof data on customer preferences. At thecore of the Total Gold rewards program(and its successor. Total Rewards, which

Instead of focusing on how

much people spent in our

casinos during a single visit,

it became clear that we

needed to focus on their

potential worth over time.

I'll describe below) was a 300-gigabytetransactional database that recordedcustomer activity at various points ofsale-slot machines, restaurants, andother retail areas in our properties.Database managers fed that informa-

tion into our enterprise data warehouse,which contained not only millions oftransactional data points about custom-ers (such as names, addresses, ages, gen-ders) but also details about their gam-bling spending and preferences. Thedatabase was a very rich repository ofcustomer information.

Slicing the DicingWhen we started digging into the data-base, one statistic stood out: Our TotalGold cardholders told us in surveys andfocus groups that they were spendingonly 36% of their annual gaming bud-gets at Harrah's. This presented anopportunity. There was a promise oftremendous upside if we could inducecustomers to spend more of their gam-ing money at Harrah's and if we couldcommunicate effectively with them. Isuggested to Satre that we might be ableto divert more of our customers' annualgaming budgets to Harrah's if we bor-rowed a page from the playbooks ofother businesses whose case studies I'dlong taught. Basically, we needed to dowhat the Starbucks and Nordstroms ofthe world had done - change the wayconsumers made decisions about ourmerchandise.

To do that, we clearly needed to sliceand dice the data finely enough thatwe could develop effective marketingprograms. Common practice calls fordefining marketing strategies apartfrom database strategies-that is, thecompany comes up with a grand mar-keting scheme and then tries to adjustthe database to its strategies. Unlikemany companies, we decided to let thedata suggest the specific marketingideas to us.

The information we found in ourdatabase indicated that a loyalty strat-egy based on same-store sales growthwould work. Same-store sales is a classicmeasurement of a simple retail-loyaltystrategy: The goal is to get a customerto visit your store regularly, just as shemight routinely visit her hairdresser andmechanic. The hairdresser and the me-chanic envelop the client in reasons tobe loyal, primarily by developing afriendly relationship. We decided to

develop just this kind of close relation-ship with the people who visit Harrah'scasinos.

Before we could persuade customersto come back time after time, however,we needed to take a hard look at themand understand how much value eachof them brought to us. We discoveredthat 26% of the gamblers who visitedHarrah's generated 82% of our revenues.We were surprised to find out who ourbest customers really were. They em-phatically were not the gold cuff-linked,limousine-riding high rollers we and oiu-competitors had fawned over for manyyears. Instead, they turned out to be for-mer teachers, doctors, bankers, and ma-chinists - middle-aged and senior adultswith discretionary time and incomewho enjoyed playing slot machines.

We also learned that thest? customerstypically did not stay in a hotel but vis-ited a casino on the way home fromwork or on a weekend night out. At thesame time, we found that our targetcustomers often responded better to anoffer of $60 of casino chips thian to a freeroom, two steak meals, and $30 worthof chips because they enjoyed the an-ticipation and excitement of gamblingitself. And we were able to developquantitative models that allowed us topredict, based on an individual's play,his or her "customer worth"-the theo-retical amount we could expect the cus-tomer to spend not just during one eve-ning but over the long term,

Suddenly, we saw how we could dif-ferentiate our brand. Understanding thelifetime value of our customers wouldbe critical to our marketing strategy. In-stead of focusing on how much peoplespent in our casinos during a single visit,it became clear that we needed to focuson their potential worth over time. Forinstance, we could see that customerswho said they were very happy with theHarrah's experience increased theirspending on gambling at Harrah's by24% per year; customers who said theywere disappointed with Harrah's de-creased their spending by 10% per year.

The best way to engage in this kind ofdata-driven marketing is to gather moreand more specific information about

MAY 2003 111

BEST PRACTICE • Diamonds in the Data Mine

customer preferences, run experimentsand analyses on the new data, and de-termine ways of appealing to players'interests. We realized that the infor-mation in our database, coupled withdecision-science tools that enabled usto predict individual customers' theo-retical value to us, would allow us tocreate marketing interventions thatprofitably addressed players' uniquepreferences. The more we appealed tothese preferences, the more money thecustomers would spend with us.

So we decided to act on a radical idea:We would reward customers for spend-ing in ways that added to their value.Most consumer businesses insist thatthey can't treat one customer differentlythan they treat another, even thoughsome customers are obviously worth

All employees are told daily:

If your service can persuade

one customer to make one

more visit a year with us,

you've had a good shift.

If you can persuade three,

you've had a great shift.

much more than others. To us, that ap-proach was fundamentally wrong, but itdidn't mean that we had to focus on therelatively small number of high rollers.Rather, we made a point of treating ourmillions of regular customers differentlydepending on their value to us.

It turned out that our customers-Iwould venture to say all customers-actually enjoy aspiring to higher levelsof achievement and reward. It's simplyhuman nature. Understanding this, wesplit our customers into three tiers:Gold, Platinum, and Diamond card-holders, based on their annual theoret-ical value. Platinum and Diamond card-holders receive greater levels of service,which adds an aspirational element tothe program. For example, our databasetold us that our best customers wantedservice quickly - they didn't want to wait

in line to park their cars, or eat in restau-rants, or check in at the front desk. Sowe decided to make a point of routingour customers into three different lines.People who weren't card-carrying Har-rah's members and Gold customersstood in lines at the reception desk orthe restaurant. Platinum customerswould stand in still shorter lines, andDiamond cardholders would rarely everhave to stand in line. This created a vis-ible differentiation in customer service.

It was essential for our customers tosee the perks that others were getting.Once we divided the lines this way,we watched as our customers did whatthey could to earn the higher-tieredcards. Every experience in our casinowas redesigned to drive customers towant to earn a higher-level card. As ittums out, marketing that appeals to cus-tomer aspiration works wonderfully.

We also set up a series of triggers in thedatabase and analyzed the customers'responses to those triggers. If, for exam-ple, we discovered that a customer whospends $1,000 per month with us hadn'tvisited us in three months, a letter ortelephone call would invite him back. Ifwe learned that he lost money duringhis last visit, we invited him back for aspecial event. Our telemarketers weretrained to listen for responses to specificoffers-a certain percentage of our cus-tomers responded positively to offers ofa steak dinner; others would respond tooffers of two free nights in the hotel.

Once entered into our database, theseresponses provided fodder for more slic-ing, dicing, and experimentation. It's im-portant to note that our database strat-egy hinged on our ability to combinedata from all of our properties, so cus-tomers could use their reward cards inmultiple locations. Combining transac-tional data from all our sites was so im-portant that we developed and ulti-mately patented the technology to do it.

We also decided to use our transac-tional data to"seH"our slot machines. Inthe past, we had no way of knowing whycustomers chose to play at certain ma-chines. Was it because of the way themachines looked? Or because other ma-chines were occupied? Or was it because

we had signs on top of them proclaim-ing the odds? Our transactional data-base told us exactly what the patterns ofplay were in our casinos. We discoveredthat at any given time, it was possible toknow which specific customers wereplaying at particular slots in Harrah'sLas Vegas and what it was about thatspecific machine that appealed to them.This knowledge allowed us to configurethe casino floor with a mix of slot ma-chines that benefited both our custom-ers and our company.

Hitting the Customer ServiceJackpotDeep data mining and decision-sciencemarketing would be worth little in driv-ing same-store sales growth were it notfor another simultaneously applied andextremely critical ingredient-an ab-solute focus on customer satisfaction.When I came to the casino business,there was an insufficient focus on cus-tomer service. We decided that great ser-vice would allow us to build-just asHome Depot, Four Seasons, and othergreat brands do in their fields-the ca-pacity to brand ourselves as the only na-tionwide consumer gaming business.

Customer service is something mostorganizations say they focus on. But, infact, they often fail to institute systemsto use customer service to reinforce loy-alty with carrots and sticks. Our datatold us that our customers want friendlyand helpful attention in addition to fastservice. We decided to linl̂ employee re-wards to customer satisfaction. Accord-ingly, we chose to measure all employeeperformance on the matrices of speedand friendliness. The better the experi-ence the guest had, the more money em-ployees stood to make. To this end, allHarrah's employees take part in a certi-fication program that trains them todeliver excellent service. From house-keepers to slot attendants, from valetsto stewards, from receptionists to chefs,all employees are told daily as they ar-rive at work: If your service can per-suade one customer to make one morevisit a year with us, you've had a goodshift. If you can persuade three, you'vehad a great shift.

112 HARVARD BUSINESS REVIEW

We implemented a bonus plan to re-ward hourly workers with extra cash forachieving improved customer satisfac-tion scores, which we culled from verydetailed customer surveys. If a prop-erty's overall rating rose 3% or more,each employee could earn $75 to $200.What has made the bonus programwork is that the reward depends on

Meeting budget at the

expense of service is a

very bad idea. If you're

not making your numbers,

you don't cut back on staff.

everyone's performance. If the valet'sscores were low but the steak housereceptionist's were high, the reception-ist would check in on the valet. Likewise,if one property received low scores andanother high ones, the general managerof the lower-scoring property might visithis colleague to find out what he coulddo to improve his property's scores.

It's important to note that we choseto measure customer satisfaction scoresindependent of a property's financialperformance. In 2002, one property hadrecord-breaking financial results, butemployees did not receive bonuses be-cause their customer service scores weremediocre. Our employees are obsessedwith their property's customer satisfac-tion scores for a good reason: In 2002,we paid $14.2 million in bonuses to non-management employees based on theirproperties'customer satisfaction scores.Since the program's inception, Harrah'shas paid out more than $43 million inbonuses.

This score-driven customer satisfac-tion measure has allowed properties -even those in troubled markets - to con-tinue to grow. Take, for example, ourcasino in Laughlin, Nevada. Despitestrong competition and a mere i% in-crease in the local market's gamingrevenues in 2002, Harrah's Laughlinrecorded a 14% gain in revenues. Why?Because its customers were loyal, thanksto great service. In fact, the employees at

Harrah's Laughlin earned the highestcustomer service scores in the company.

Our experience with customer servicehas shown us that meeting budget atthe expense of service is a very bad idea.If you're not mal<ing your numbers, youdon't cut back on staff. In fact, just thereverse: The better the experience aguest has and the more attentive youare to him, the more money you'll make.For Harrah's, good customer service isnot a matter of an isolated incident ortwo but of daily routine. When he goesoff duty, the Laughlin general managertells employees to call him at home-anytime, day or night-whenever theysee five people waiting in any line. To us,this is living proof that our same-storesales growth in tough markets has beendriven by sustained attention to greatcustomer service.

The Winning HandEverything we do to market Harrah's isframed in terms of players' decisions tovisit, or not visit, one of our casinos. Onemeasure of the effectiveness of our strat-egy is that many of our competitors haveadopted similar programs after viewingour company's performance over thepast few years. But our competitors con-tinue to focus largely on facilities, whilewe keep combining improved facilitieswith breakthroughs in marketing andcustomer service. We maintain our com-petitive advantage by using our humancapital and technology systems to get toknow our customers better.

Harrah's will keep adding excitementand benefits to the Total Rewards pro-gram, widening its scope across gaming-related activities. And we'll keep en-hancing the benefits that players getfrom consolidating their gaming withinour brands.

Let the neighbors lure tourists withknights on horseback, fiery volcanoes,pirate ships, and mini-Manhattans. We'lljust keep refining what we're alreadypretty good at: drilling into our data andmaking sure our regular customers aremore than satisfied. ^

Wednesday, May 7, 200312:00 - 1:30 p.m. Eastern Time

THE SUCCESS ATTRIBUTEDto Six Sigma at companies such asGE Is widely known. But not everyorganization has similar results.

How can your company unlock thetransformative power of Six Sigma?How can you embed Six Sigma inyour corporate DNA? What is therelationship between leadership

and Six Sigma performance?

Join Six Sigma expertSubir Chowdhury, CEO of the

American Supplier InstituteConsulting Group and author of

the bestselling The Power ofSix Sigma, and leadershipauthority Noel Tichy of the

University of Michigan BusinessSchool for a 90-minute interactive

virtual seminar in your office orconference room.

For details, visitvvww.asiusa.com

orcaii 1-800-462-4500

Reprint R0305H; HBR OnPoint 3647To order, see page 131.

MAY 2003