HEAD GAMES
Businesses are deploying analytical software to get a better fix on customer behavior.
By John Goff

Casinos may be glittering, neon-bathed temples to the cruel goddess of fortune, but when it comes to wringing money from wallets, casino operators leave precious little to chance.

Consider Harrah’s Entertainment, a Las Vegas–based gaming specialist with 26 hotel casinos in the US. Last year, Harrah’s generated US$4.3 billion in sales, much of it from gambling operations. The key to Harrah’s green machine, say observers, is the company’s intimate understanding of its customers. Harrah’s hotels, for example, had a remarkable 95 percent occupancy rate in 2003. More remarkable: the company turns down twice as many requests for reservations as it accepts, which is why it continues to invest in new hotel rooms.

“It’s not about filling each room,” explains David Norton, senior vice president of
relationship marketing atHarrah’s. “It’s about maxing out the profit from each room.”

Toward that end, Harrah’s launched a hotel revenue management system (RMS) in 2001, enabling it to optimize the profitability of its hotel rooms through a combination of gaming revenue and room rate. RMS forecasts occupancy at a detailed customer-segment level based on historical and projected trends, and makes a decision in real time about whether a customer should get into the hotel and at what room rate, based on the customer’s profile in the data warehouse, which is based on NCR Teradata technology. The customer will get a consistent answer whether booking over the phone or online at harrahs.com.

Patrons who play a lot get sweetheart deals from Harrah’s. They also tend to get lots of mail from Harrah’s. “About 75 percent of our revenue comes from direct-
marketing offers,” notes Norton. “If we didn’t do it, our revenues would tank.”

Few companies are as zealous about communicating with customers as Harrah’s, and judging from the success of Total Rewards, the company’s loyalty program, customers appreciate the contact. In fact, Harrah’s is currently looking into communicating with customers while they are playing in the casino. But many companies are likewise considering the use of customer relationship management (CRM) technology to wring more money from their customers’ wallets. First developed in the early 1990s by Siebel Systems as a management tool for sales personnel, CRM has since morphed to include campaign-management applications, call-center software, and customer self-service programs.

Despite years of declining interest in CRM, and despite its miserable track record – just 16 percent of projects result in a positive ROI, according to Boston-based AMR Research – spending on CRM products is now on the rise. The hottest segment in the market is customer analytics – tools that dissect consumer-buying patterns, suss out preferences, and predict future behavior. AMR reckons that sales of business-intelligence/ analytics products will top US$9 billion this year, up from US$7.7 billion in 2001.

Why this surging interest in analytics? Analysts say frustration over earlier CRM projects may be fueling current sales of CRM analytics products. After funneling large amounts of capital into call centers – centers that are now faster but not better – executives appear keen to get something for their CRM money. “Companies want to know how they can turn these cost centers into revenues,” notes Stan Martin, CEO of US-based Adroit Consulting. Mining the prodigious amounts of data generated by call centers and other points of customer contact may be one way.

Not surprisingly, software vendors have been quick to jump on the analytics bandwagon (see “The Vendor Landscape,” page 40). Some vendors – prominently Business Objects, Hyperion, and Cognos – are flogging programs that collect and measure sales data. Others, such as business-software giants Oracle, PeopleSoft, SAP, and Siebel, offer software that analyzes buying trends. Still others (including Teradata and SAS) market predictive-modeling packages. And a number of vendors sell applications designed to group customers by categories, including profit potential.

Eventually, analysts believe, there will be a blurring of the lines, with software makers offering analytics products that measure, analyze, and cluster. But CFOs, some of whom saw elaborate call-center initiatives go awry, will take a bit of convincing. Says Rick McMahon, CFO of Sunstar Butler, a Chicago-based oral-care products company that is contemplating buying an analytics program: “It’s just way too much money and time to end up being a toy.”

Science of selling

Finance chiefs like McMahon have been down this road before. During the go-go days of corporate call-center spending, vendors hurriedly rolled out different customer-service applications. The goal was to provide a 360-degree view of a consumer, but more often than not, the view was less than ideal.

Case in point: Jonathan Wu, senior principal at US- based professional-services firm Knightsbridge Solutions, recalls one client that operated six different applications that interfaced with customers – but not with one another. The siloed systems generated a substantial amount of conflicting and duplicate data. No big surprise, then, that the company’s management had a slightly overblown opinion of how business was going. “They thought they were growing by 23,000 customers per month,” remembers Wu. “It wasn’t even close.”

Such stories have not exactly burnished the reputation of CRM software vendors. Nevertheless, the promise of analytics – better selling through science – appears to be gaining converts. Indeed, some companies are already engaged in predictive modeling and segmenting.

With predictive modeling, companies attempt to forecast future customer behavior based on analytic models. Segmenting, on the other hand, involves grouping customers by common traits or behavior patterns; clustering is one common analytic technique to help achieve this. Generally, businesses segment customers into groups to help them devise the most cost-effective way to market and to service those groups.

Segmenting is not limited to existing customers, however. At California-based Volvo Cars of North America, Phil Bienert, manager of the automaker’s CRM & E-business group, says his department is currently in the middle of a segmentation project involving prospective customers. According to Bienert, Volvo is breaking current customers into segments, and then comparing the patterns of those groups with those of prospective buyers. The patterns can be obvious – customers moving up the auto food chain, for example, from compacts to midsize cars to SUVs – or hidden, the kind that companies need analytics to uncover.

The goal is to identify behavior that indicates a propensity for buying a Volvo down the road. “You can apply these owner characteristics to hand-raisers (those who request information about Volvo products) and cluster them,” explains Bienert. “Then you can prequalify people who haven’t even entered into communications with the company.” (Like many large companies, Volvo buys consumer data from third parties.)

Should you dump customers?

Of course, not all hand-raisers will prove to be valuable customers. Indeed, some executives argue that not all customers are valuable customers. At US electronic and industrial communications products maker Woodhead Industries, CFO and vice president of finance Robert Fisher says management’s thinking in the past has been that any customer is a good one. “We’ll sign up anybody who will sell our products,” he says. “That may not be so smart.”

To get a little smarter, Woodhead has started developing a framework for lead management and contact information to work with global customers like DaimlerChrysler and Ford Motor. The company is also implementing a business-intelligence platform from PeopleSoft. Fisher says the technology will enable Woodhead management to see instantaneously what the company is shipping, by customer, product line, location, and the like. Further, sales managers will be able to group customers by gross margins. Says Fisher: “I want to identify customers who I want to spend more time with and the ones I want to dump.”

He’s not alone. Ditching costly customers has become something of a corporate mantra in the past few years. But some consultants say relying on clustering to deep-six a segment of customers can be a tricky business. In fact, some rail against the practice. “Rarely is it a good idea to dump a customer,” insists Laura Preslan, research director at AMR. “The cost of acquiring a new one is so high.”

What’s more, profit profiles (which are usually based on overhead and other support costs) can be way off. Warns Gareth Herschel, research director at research firm Gartner: “You may end up kissing off an entire segment of valuable customers simply because you misfigured the depreciation of a printer in Poughkeepsie.”

Analysts also point out that today’s costly customer could turn out to be tomorrow’s cash cow. Herschel recalls how, during the early 1990s, many companies were eager to ditch customers that placed a lot of calls to customer-service centers. But then the internet came along, making it cheaper to service those customers and providing a lot of low-cost cross-sell and up-sell opportunities. “You were desperate to get rid of a customer,” he says. “Now, you’re not.”

In other words, customers and their circumstances change. That’s why analyzing customer information remains an imperfect – and never-ending – quest. A case in point: a director of analytics at an internet service provider (ISP) points out that users with the highest risk of canceling their service used to be those who simply didn’t find the internet interesting. Based on that data, the ISP might have considered launching such things as streaming video and music. Over time, however, the profile of likely defectors has changed. Nowadays, the customers most likely to break their service contracts are those who have slow-processing computers. Launching more entertainment features – those that require fast processors and tons of bandwidth – will do little to ease their pain.

For CFOs, the lesson is clear. Analytics tools are just that – tools, not cure-alls. “[Things] don’t change simply because you open a box and pull software out,” says Sunstar Butler’s McMahon. “You’re still the same company.”

John Goff is technology editor of CFO in the US.