Follow these 10 steps and you'll have information that really means something, rather than just have a lot of data that confuses more than it informs.
For the rest of the April 2003 issue of CRM magazine please click here
You have a substantial customer database. You rejoice in its depth of information and in the significant effort made to collect the data. We salute you. But we must also inform you that it's not enough. Without the right tools and strategies to make the best use of that data, it will just sit and get stale. Like that unidentifiable greenish substance in a container in the back of the fridge, what was once grade A not only becomes unusable, but the wasted investment could cost you dearly.
So how do you ensure that your database stays fresh and pays for itself and more? Just follow these 10 steps and you'll have information that really means something, rather than just have a lot of data that confuses more than it informs.
1 Ensure Data Quality
Data quality means more than simply knowing who is a Mr., Mrs., or Ms. It means being alert for stale entries, bogus addresses, and duplicate records. Poor data quality means higher marketing costs (such as mailers that end up in the dead letter office), and the potential to seriously misunderstand the true profile of an important customer.
"If one customer record is 'John Deere Inc.' and another is 'JDEERE', and the querying tool doesn't recognize the two as one and the same, then one is worth $1 million in revenue and the other is worth $200,000," says Leonard DuBois, vice president of marketing for Trillium Software. "You want to see that it's one company worth $1.2 million to you. That's the value of a data quality solution."
Every night Harrah's Entertainment runs a database process with manual inspection to merge any duplicate customer records that might be created unwittingly in its rewards program. The process saves on marketing dollars and ensures that customers always receive the credit and treatment they have earned.
2 Measure Success on Metrics That Matter
The flexibility of modern analysis and reporting tools is as much a blessing as it is a curse, as hundreds of meaningless reports can be created at the click of a mouse. Avoid this problem before it starts by working from hypothesis to test to conclusion, rather than by simply churning through instinctive what-if scenarios until you stumble across a pattern that provides a satisfactory explanation to a business problem.
This means you need to spend ample time researching and planning to identify how a campaign is intended to perform, or how a certain customer segment is expected to perform over a given period of time. Define how that success can be evaluated; and then stick with a plan of analysis from start to finish, rather than fudge the formula halfway to the target. And while the targets you select can certainly be more varied than the obvious "average customer revenue," avoid the urge to go overboard by trying to score dozens of metrics with no understanding of how they relate to one another.
Along with selecting analysis that makes sense, select a reasonable number of items to analyze at any given time. In a previous position managing a call center, Witness Systems' principal market consultant Oscar Alban found that his technical operations staff spent time and money needlessly running hundreds of reports every night that nobody read. "I looked at three of them, but nobody told MIS not to [run the rest]." That time and effort could be better spent on new projects or refining existing measurements.
3 Enable Users to Get the Insight They Need
This means, get beyond prepackaged reports and provide tools to the users that are relevant to varying levels of user sophistication, from sales staff to analysts. It also means that data should not be static--it should lead to an immediate course of action, by tying analytical results with workflow, for example. "Go beyond telling me where I am. Tell me what I should do about the fact that I have 250 different leads in my SFA application--like what I should do with those leads. Or with the customer I have on the phone right now in terms of driving them to a problem resolution or a cross-sell," says Brad Wilson, vice president of CRM marketing for PeopleSoft.
Putting analytical tools in everyone's hands is another way to deliver insight from data. And doing so can be quite liberating. "You don't have to package data in a way that assumes what [the user] is going to do with the data," says Ralph Doran, vice president of research and development at Databeacon.
In fact, to help its customers get exactly the information they need, the way they need it, Macdonald & Associates, a private equity research firm based in Toronto, uses software from Databeacon to provide sophisticated modeling and analysis tools directly to those customers via a Web portal. "It effectively feels like a custom research shop in the hands of our users," says company president, Mary Macdonald.
4 Unify Data Across Channels
There is no single more-tired analogy in decrying the problems of modern business than the "tragedy of the stovepipes." But until it goes away, companies must find and use tools that allow them to recognize customer behavior at multiple touch points as belonging to one and the same relationship, or it is impossible to get a true sense of each customer's value and their pain and pleasure points.
Bringing multichannel data together also keeps business users from pestering analysts or cross-divisional experts for insight into business trends and issues outside their bailiwick. "By democratizing the data, you make [analysts'] jobs easier, because people will do the easy stuff themselves," says Naras Eechambadi, CEO of marketing services firm Quaero.
Multichannel also means watching industry trends and making them a part of your overall business insight. For instance, SAP reports working with pharmaceutical companies that pull segmented national prescription data from third-party sources, obtaining more comprehensive results than they could get from locally aggregated information. They then use that combined data to create relevant compensation plans for their sales forces.
5 Establish Meaningful Customer Segments
Just as you can run hundreds of meaningless reports, you can establish just as many vaporous customer segments. Don't let assumptions guide the definition of customer segments--design a hypothesis of what factors make customers different, and keep in mind that it may take several iterative passes examined with a fine-tooth comb to reach a final decision on what constitutes the best segments.
Marketing consultancy ANALYTICi worked with a video-game maker for the European launch of a new console and ran a campaign tying new system sales to an auto-racing game. Using a database of one million customer records, they found that identified racing enthusiasts in the under-15 and over-25 age brackets were most likely to buy into a bundle deal. The video-game company made that offer to 200,000 such customers, along with an extra 10 percent in a control group comprising 15- to 25-year-old non--race fans. The campaign resulted in a verifiable 12 percent boost in revenue for the title from the target group (data was unavailable at press time for the control group).
Remember that as customers, retailers are not just minidistributors, and that consumers are motivated by entirely different concerns than are retailers and distributors, so what the data tells you about a regional distributor may not map closely to the customers ultimately served by that distributor. For example, consumers may underperform on manufacturers expectations not because of a lack of interest but because of weakness in the demand chain. Learn to tell the difference.
6 Encourage Customer Growth
Customers grow given the right amount of attention and time. Once you understand your high-value profiles, look for customers in the lower tiers who, aside from their contribution to the bottom line, most closely resemble those high-value customers, and uncover whether it is ability or willingness that keeps them from closing the gap--then take action accordingly.
Harrah's, for example, has been working with the same customer segmentation rules for the past three years. It continually tests and adjusts the performance of offers of cash, food, and hotel discounts against customers it sees as having growth potential based on proximity, propensity to gamble, and any other information the customer might have volunteered.
Harrah's also realizes that customers who are loyal, but not as yet high-value, may become high-value over time. Generating offers at Internet speed doesn't necessarily double wallet share or create insatiable demand at that same speed. "People don't become gamers overnight, and you have to have disposable time and money," says David Norton, senior vice president of relationship marketing for Harrah's.
Another key to encouraging growth is to stop decline. Use automated database triggers to alert the organization when customers start to drop off, possibly launching a campaign to stay on the customer's radar. More important, use analysis to recognize customers who are likely to churn; then employ appropriate retention strategies to regain their loyalty before they head to the competition.
7 Take the Bad With the Good
Don't try to learn only from good news and nice numbers like profit growth. Studying customer losses, reports of negative experiences, and what makes the customer say no can be even more telling than a glowing quarterly sales report. The answers may provide far more value than the good news about what the business isn't doing right. For example, "there may be a situation where upsells have not increased to the desired level, because agents are not handling objections properly," Witness Systems' Alban says.
In fact, analyzing the bad news should happen regularly. The benefit could be increased customer satisfaction. "An overall campaign might be successful and meet our hurdles, but five segments didn't perform to the level we want, so we'll go in and analyze those and make changes," Harrah's Norton says. "Don't just look at [negative data] every six months. [By then] we've lost the customer."
8 Model and Predict Profitable Loyalty and Motivations
"Understand that your most loyal customers may not always be your most profitable--they may be loyal because there are factors about how you're dealing with them that may not be the most profitable to you," says Stephen Horne, president and CEO of ANALYTICi.
Getting to a true understanding means going beyond the easy numbers. "You need to convert simple measures of responses, transactions, and revenue to measures that are meaningful to business, such as margins, profitability, share, and customer retention," Quaero's Eechambadi says.
If data shows that customers are motivated to behave unprofitably, don't snap to punish them with disincentives and fees. First look for the underlying reasons for their action, as one E.piphany customer did. This travel agency noticed an increase in profit-sapping phone transactions to cancel rental car reservations. An audit of the agency's Web site uncovered that the cancellation interface was too difficult to find, so the problem was corrected and overall customer interaction costs settled down.
9 Make the Right Call (or Email, or Flyer, or...)
Your database should represent a world of possibilities for communicating with customers, prospects, and defectors at a moment's notice, fueled by triggered campaigns. But make sure your database reflects, in bright lights, how those customers want to hear from you.
"You should constantly have in the data warehouse an event-tracker looking for customer events, and immediately get back to the customer with a message, an offer, a banner. [It's] a process of communication that's relevant to what the customer is doing," says Ron Swift, vice president of strategic customer relationships for Teradata. That means more than just making a real-time recommendation, however. It means watching for life-cycle events, it means watching for customers to transition to a new profile, and it means staying ahead of potential lost opportunities. "Don't stop marketing to former customers. Win-back customers don't necessarily churn [again]," Swift adds.
In addition to event-based communication use data analysis to find your most loyal customers, then check in with them. In the experience of Mike Coakley, vice president of marketing technology products for customer contact outsourcer Epsilon, truly successful companies know how to ask questions and listen to the answers. "Find out what's important to these customers, your real loyal customers who spend money all the time through various channels. Ask them, 'Why do you like doing business with our company?'"
10 Keep It Safe
Finally, keep your data safe from harm and misuse. While aggregating data and opening up analysis is attractive from a business perspective, it does offer the potential for abuse. At Harrah's, "Only the serious marketers can extract data, and they go through rigorous training and have to get the authority to run a [mailing list]," Norton says. Backup is, as ever, very important. "You should have a standard disaster-
recovery plan, have redundancy, and be able to get to [your data]," says Todd Bixby, managing director at BearingPoint.
As a small firm with limited IT resources, Macdonald & Associates pays an outsourcer for complete hosting and backup, including a service contract that calls for well over 99 percent uptime. Medium-size companies should consider the use of virtual server technology, which can redeploy disk images on virtually any hardware, rather than having to be perfectly rematched to the same equipment in the event of failure. Larger firms may need to engage multiple outsourcers or use a series of offsite backup and potentially even extra equipment stores to ensure that operations cannot be more than minimally disrupted.
Jason Compton is a freelance journalist based in Evanston, IL
How Nortel Gets Data to the Right Users at the Right Time
Nortel Networks has 14 specialty data marts, 3,000 analytics users, 19,000 CRM users, and 1 boss to oversee it all. Marykay Wells is the IS leader for value chain solutions at Nortel, and is the corporate leader both for the CRM implementation and the massive corporate data warehouse that was created when Nortel merged 150 customer-related applications into a single system.
"We wanted to consolidate the sales view and the marketing view, and get that information out so we could make good business decisions," Wells says. Nortel's implementation started in mid-2002, and the major phases require about 18 months to complete. Along the way, the enormity of the task began to sink in. "We had moved our systems to Clarify, but had underestimated the amount of reporting we needed to do, both inside the application and outside," Wells says. It wasn't until the CRM upgrade had begun that she and others came to understand that their success would not lie in integrating the legacy customer interaction and analysis applications, but in making the data those applications supported more powerful and accessible. "We didn't realize that the real value was getting the data out and being able to report on it, so we really had to expand our global data warehouse solution to get that value."
In the end Nortel's success, aside from having brought together so many applications and data sources that worked at all, is that the data warehouse project has stayed true to the vision of providing the right access to the right data for the right employees. "We had to put some thought into how to segregate the information and really secure it," she says. "A person viewing certain information doesn't need to look at sales opportunities, and executives wouldn't use the CRM application, but would get their information from a reporting screen."
The integrated, role-appropriate interfaces save time and provide mission-critical information to the right business users. "We consolidated the sales view and marketing view all in one place, and get that information out so we could make good business decisions," Wells says. Nortel's integrated data warehouse has led to roughly $10 million in savings through systems decommissioning, and helped preserve accurate and timely insight and analysis as the company goes through substantial workforce reductions. --J.C.