EBay, the online auction service, had a problem earlier this year: Although visitors to its “my eBay” Web pages were able to specify the categories they wanted to see, they still had scroll through pages of listings to view new items – 250,000 of which are added each day.
Realizing that it doesn’t pay to waste customer time, eBay decided to upgrade its system. In April, it launched Personal Shopper, a free, customized search service that relies on E-Business Minder software from NetMind.
The software detects new items and then, based on choices contained in eBay’s shopper database, triggers an e-mail notifying the customer of specific items of interest, creating a “personal valet service,” says Brian Swette, senior vice president of marketing at eBay.
Thus, the firm identifies and differentiates its customers, interacts with them, and customizes what it sends them – all four steps of the one-to-one process – in real time. “The key is to convert site visitors into repeat buyers,” says NetMind senior vice president Kelsey Phipps.
Welcome to the new database marketing. Although based on many of the same premises as the old, it goes beyond it in ways that were not even imagined until recent years.
Because much of it is driven by the Web, there is a higher degree of interactivity with the customer. Firms like eBay take action, sometimes instantly, based on the stored memory of individual transactions and interactions, resulting in an unprecedented level of customization.
Although a Forrester Research analyst predicted last spring that spending on data mining software will double to $3.56 billion in 2001, an executive survey by Forrester showed that only a minority of firms are now mining their databases, according to The Wall Street Journal. Even fewer may be using their databases to pursue instant dialog with customers. But anecdotal evidence shows that eBay is hardly the only firm to try it.
Amazon.com, the Seattle-based online bookseller that recently began competing with eBay in the auctions arena, also uses database-driven e-mail to notify its 11 million customers of new offerings, for example, when an author whose books they’ve previously purchased hits the shelves with a new title.
“With millions of customers, we can provide powerful recommendations based on knowledge of what similar buyers have purchased,” says Amazon.com’s Paul Capelli.
The key to the learning relationship for Amazon.com is the company’s ability to analyze purchase patterns and cross-reference based on what similar buyers have purchased. Thus, it can customize its offerings on an individual basis.
Then there’s Replacements Ltd., of Greensboro, NC, an online service (www. replacements.com) that helps customers locate china, silver, crystal and collectibles to replace missing or broken pieces (it has an inventory of more than 6 million items).
The firm uses advanced database engines that let it match vast amounts of customer and inventory data. Among other things, it uses customized e-mail to cross-sell, for example, by letting certain customers know a gravy boat that matches their original set is now available for purchase.
It also uses this data for more traditional purposes. When new items arrive, detailed information is cataloged in a database, and customers are notified by mail. To place an order, the customer simply makes a telephone call, references the mailing piece and the item is shipped.
For customers who want preferential treatment, Replacements also has a “Call Collect” program. This service informs buyers who are willing to accept a collect telephone call when specific pieces in a pattern become available – before written notification is sent to other customers who have requested the same item.
BUSINESS-TO-BUSINESS
Not all of these state-of-the-art applications are directed at consumers. Dell Computer, the world’s leading direct seller of computers, is locking in its high-volume accounts by creating personalized Web pages for them.
Using Dell’s Premier Pages, MIS personnel from the corporate, government and educational sectors can configure systems based on their firms’ approved specifications, which Dell collects and stores in its database.
Purchasers see only those products they are authorized to use in building systems. When they have done the initial work, a quote is sent to the buyer, who can either reject it or forward it back to Dell and complete the transaction. Built into Premier Pages are multiple levels of user access that can be defined by an organization.
To date, Dell has created more than 24,000 Premier Pages, and it now conducts more than 50% of its online business through this channel, reports Dell’s Andy North. Once a customer signs up, says North, it becomes easier to buy from Dell and harder to buy from other vendors.
That’s not the only benefit.
Having multiple points of contact within a customer’s organization greatly enhances Dell’s value and also makes the relationship far less price sensitive.
CUSTOMER DIALOG
But none of this is easy. “Without a doubt transactions through a non-human channel require a more systematic approach to gathering and using customer information,” says Anthony M. Marsella, director of marketing intelligence for IBM Worldwide Software Group, in Somers, NY, speaking at the recent Marketing to Business Conference in San Francisco.
It starts with getting to know the customer.
Traditional DMers went through at least two stages of the one-to-one process: They identified their customers, and differentiated them as well as they were able – even if it only meant appending geodemographic data from a compiled database. But they couldn’t get down to the one-to-one level because they didn’t see communication as a two-way street.
“With database marketing, there is huge potential, but with that potential comes an unspoken responsibility to engage in dialog,” says Rick Barlow, president of Frequency Marketing, Cincinnati, OH. “We know that if we can position customers to think of themselves as members or insiders, they begin to open up at a staggering rate.”
Some companies capture customer information by tracking clickstreams on the Web and recording items placed in shopping carts – thus, the dialog is in a sense involuntary. But many consumers are willing to provide information about themselves for a quid pro quo. “They believe that by investing in the relationship in exchange for providing information, the marketer will provide rewards,” Barlow says.
This seems to be the logic behind many Web start-ups.
For example, ImproveNet, an online competitor to Home Depot, offers an array of services for the homeowner contemplating a remodeling project. Realizing that many consumers are wary of contractors, the firm maintains a database of 600,000 pre-screened contractors, architects and designers whose references and legal and credit histories have been checked by the company’s specialists.
To locate a local tradesmen, a homeowner simply logs on to the site, describes the project, and ImproveNet takes it from there, contacting 40 to 80 approved contractors within a 10 mile radius of the customer’s ZIP code. The contractor pays a lead fee of $6 to $10 and, if awarded the job, gives ImproveNet roughly 1% of the revenue.
But the service doesn’t end there. “Once we’ve matched the homeowner with the contractor, we don’t let the consumer go,” says Bill Crosby, vice president of editorial at ImproveNet, based in Redwood City, CA. “A personal project advisor is assigned to answer questions along the way, and in the end, to make sure it was a good experience.”
What does ImproveNet get out of it, besides the 1%? It gets to load up its database with all data relevant to the project, and use it to sell advertising to manufacturers. Although it never provides names directly to advertisers, it offers three other options: online advertising on ImproveNet’s Web site; direct e-mail communication with prospects who have expressed interest in a product category; and targeted offers inserted in customers’ Personal Project folders.
“Homeowners are self-selecting,” says Crosby. “They are coming to us, raising their hands and telling us that they are interested in refrigerators, faucets and cabinetry, and we can assist by providing information on those items.”
In July, ImproveNet processed over 16,000 projects nationwide. And since its debut in 1996, the company has grown to 80 employees and had its hands on over $2 billion in projects.
Another home-related service is offered by Ruhl & Ruhl Realtors (www. ruhl-ruhl.com), located in Bettendorf, IA. Logging onto Ruhl’s site, a visitor sees an icon that asks, “Would you like to talk to a NetRep?” (The firm uses NetRep software.) If the response is yes, a notice is sent to the business development department at Ruhl’s office, and within minutes, a representative from the company initiates an online dialog with the visitor.
“We’re able to answer questions about buying and selling homes as well as provide information about financing, all in real time,” says Shelly McVietty, client relations manager for Ruhl & Ruhl. “If someone wants to know if they can afford a home,” she explains, “we have a financial services rep on site who can get loans pre-approved. With NetRep, we’ve pre-approved buyers in 20 minutes.”
Once the NetRep is contacted, the lead is then entered into a referral database and an agent is assigned. This process has snared some big fish.
Based on ISP addresses, Ruhl’s marketing staff noticed significant activity on the site from John Deere. Instead of waiting for a call, a relocation expert from Ruhl contacted a vice president at Deere to determine if there were any transfers or changes in the works. There were.
ONE-TO-ONE OFFLINE
Not all applications are online – smart scripts used in telemarketing are based on the same premise. For example, in attempting to sell new subscribers and upgrade current ones, the phone reps for a large daily newspaper access a central database, and immediately identify the date and reason for all previous contacts, including complaints, past delivery problems and subscription history, according to Sue Hardin, manager of information and database marketing at Yeck Brothers Co. in Dayton, OH.
With this bank of information, the phone rep can initiate a conversation with the prospect by referencing their past history with the company, and then employ different pitches depending on the subscriber. By being able to connect more personally with each customer, the paper’s telemarketing department has been able to increase its activation rates significantly, according to Hardin. “Rather than constantly sending one-way communication,” she says, “a dialog gives us the tools we need to address the customer’s needs during the next conversation.”
American Express, which has 22.5 million U.S.-based card members, pursues a wide range of personalized selling strategies, all driven by the company’s database. If a retailer opens in a new location, for example, American Express can offer discounts to consumers who live within a certain geographic range of the new store whenever they use the card. In addition, the company sells targeted advertising printed directly on customers’ monthly statements, based on activity with a given merchant.
The database can also drive both outbound and inbound telemarketing programs. A caller whose record indicates he or she is a frequent traveler may be offered baggage insurance. If a cardholder calls to announce an address change, the telephone rep is prompted to ask if the caller wants to be enrolled in the company’s credit card registry.
Likewise, frequent Delta flyers are invited to apply for the co-branded American Express/Delta credit card. Do customers mind these cross-sell efforts? Hardly. “We know it’s a convenient time because the customer has contacted us,” says American Express spokesperson Gail Wasserman. “And we’re using knowledge based on information we have about our customers’ buying habits to make relevant offers.”
Some firms are able to customize in person. Harrah’s Entertainment Inc. uses an NCR Teradata database warehouse system to track the habits of the more than 14.5 million guests who eat, sleep and wager at the company’s 18 casinos. It uses the data, among other things, to determine which rewards – such as free entertainment, food vouchers and accommodations – to offer repeat customers in its Total Gold program. The customers simply swipe a card for every transaction, and magnetic strip technology does the rest.
Rewards are based on more than simple volume.
For example, the company might award hotel vouchers to an out-of-state guest, while free show tickets might be deemed more appropriate for patrons whose visit is a day trip. Harrah’s also tries to encourage cross-market play – that is, visits to different Harrah’s facilities. Since the Total Gold program started two years ago, the firm has increased cross-market revenues to 12% of total gaming revenue, up from 8% a year ago. And according to John Boushy, senior vice president, Harrah’s information technology and marketing systems, that is due in large part to the fact that, “We are able to keep track of millions of customer activities and we are also able to analyze and predict the true value of each customer.”
Not all database applications even involve customers in the traditional sense. Alliance Capital, the largest publicly traded asset manager in the U.S., recently streamlined its system for providing news and information to 5,400 of its highest-volume brokers. (These “customers” account for 80% of its sales volume).
Alliance had Data Communique, New York City, design and build a database-driven fulfillment system and Web site (www.e-lert.net). Data Communique acts as a portal for incoming financials that are loaded into a database. The top ten holdings are calculated along with performance numbers, and this data is used to create customized fact sheets for the broker/dealers who work with different funds.
Prior to building the site, Alliance sent a business reply mailer to each of these individuals, outlining the plan and inviting them to provide personal information. Brokers entered their preferred funds, market focus, and preferences for delivery of sales collateral, says Donna Lamback, Alliance’s vice president for print production and fulfillment.
OBSTACLES
But there are problems aplenty awaiting unsuspecting companies. One is channel conflict, particularly in the business-to-business arena. The more businesses move toward an Internet-based sales model, the more they risk alienating salespeople who are still the customer’s ultimate point of contact if anything goes wrong.
Dell says it has solved this dilemma by empowering its salespeople to focus on strategic account management – sitting down with customers to determine their satisfaction levels and future needs, rather than trying to get them to buy that extra unit – and by creating a compensation structure that allows them to do it. “What our sales force has found is that they’re spending less time on day-to-day transactional activities and more time on high-level relationship building,” explains Diane Block, online manager for relationship customers at Dell.
Another issue is how the database is managed in-house. True one-to-one marketers would build enterprise-wide systems, which provide a global view of the customer and are accessible by all business units, from the marketing department to accounting. For example, IBM worldwide Software Group is now rolling out in the U.S. a database marketing system that now covers 16 European nations. The U.S. system, comprised of 150 internal databases, includes data gathered by both back-end systems and Web sites.
Why go to the trouble? Because the older processes generated incomplete or incompatible data, requiring a significant amount of manual reprocessing of data, according to IBM’s Marsella. Big Blue no longer uses its database only to track product sales leads and field sales force activity; it now uses it to manage all communications with customers.
As data is collected in the new system, it is used to fuel self-adjusting predictive modeling programs. These are used for interacting with customers in a continuous one-to-one dialog.
But these are not the only data-related problems – especially for firms operating on the Web. “In many cases…they are finding that the potential reach remains far ahead of their actual grasp,” wrote Lisa Bransten in The Wall Street Journal last June.
That may be because existing database management software is unable to cope with the amount of data available. “About 90% of the data that’s collected on the Web is unstructured,” said Marsella. “Little data management is done.”
IBM hopes to remedy that by developing new software to manage database content. Marsella predicts that data management and tracking in general will improve as new online media develop.
To hear the consumer press tell it, these tools have already been perfected.
“Today the Web has seen an explosion of marketing organizations that have used the infrastructure in unforeseen ways to build remarkably invasive profiles of Internet surfers,” John Markoff wrote in the New York Times this past summer.
“By combining vast databases with data mining software to form something resembling a vast vacuum cleaner, these new automated systems are rapidly tracking as well as predicting human behavior,” Markoff continued. “What advertisers have been doing on a mass scale for decades is now done one-on-one.”
Given this kind of press, one-to-one marketers would do well to post their privacy policies on their Web sites and honor customer wishes about use of the data. That includes allowing them to opt out of the process.
But there is another side to this – that far from creating paranoia, collecting of information on a person often raises expectations.
“The marketer had damned well better use the information because if there is no action, the company’s credibility erodes,” Barlow says. That, in turn, leads to the worst problem of all: “If a firm invites a customer to invest and nothing is done with the information, even something as simple as a thank you, there is an implied invitation for the client to withdraw from the relationship.”
In many respects, one-to-one marketing executives are a lot like the gold miners who rushed to California 150 years ago. They dig mines, sift through streams, and try to uncover the means of attracting and retaining the most valuable customers.
However, rather than toiling with picks and pans, today’s miners are using computer technologies to sift through databases to learn more about individual customers, discern patterns, and use this knowledge to gain a competitive advantage. For an enterprise to implement one-to-one marketing programs, of course, the ability to take advantage of comprehensive customer databases is the only way to develop and maintain an ongoing relationship.
Toward that end, the approaches to accessing, processing and leveraging information have ebbed and flowed along the same lines as the evolution of computer technology.
During the heyday of mainframe computing, virtually all customer and corporate data was stored centrally, giving rise to data warehouses. Like their physical counterparts, data warehouses usually contain data about the entire company, including products and services, customers, vendors, transactions, accounting, financials and manufacturing.
With the advent of minicomputers and then PCs, there was a trend toward distributed computing, which resulted in data marts. Data marts are most frequently used by sales and marketing folks to figure out what’s selling, what’s not, and which products are the most and least profitable. The flip side of this concerns customers, with marts used to assess clients’ loyalty, price sensitivity, as well as identify cross-selling opportunities and test promotions. Given the results orientation of users, data marts tend toward summaries that identify patterns and general trends rather than showing all the data points.
And now, as networked computing gains momentum, the notion of centralized storage is back in vogue, though with a bit of a twist. The current trend in data mining is toward a hub-and-spoke architecture in which departmental data marts are anchored to a central data warehouse. In fact, no one interviewed for this article advocated the notion of using data marts that operate separate and distinct from the warehouse.
Build it and they will use it
The essence of a data warehouse is to build a single, enterprise-wide database that can be accessed by any permitted user for any business purpose. When properly constructed, a data warehouse contains what is often referred to as “one version of the truth,” meaning that there is a consistent nomenclature, a single set of metrics, and filters to prevent duplicate or “dirty” data from diminishing the accuracy and validity of the information. Another characteristic of the data warehouse is that it represents a historical view of every transaction between the enterprise with both customers and vendors.
>From a one-to-one viewpoint, the sheer size of a data warehouse offers >significant benefits as well as some major drawbacks. First, by building a >massive database, the conclusions reached through analysis – whether by >OLAP or data mining techniques – are more likely to be valid, just as with >any statistical sampling. “If you consolidate the data, you have the >opportunity to mine it to generate more accurate analyses,” notes David >Menninger, vice president of product marketing and data management for the >data mining division of Oracle.
Further, by recording more information about customers and their habits and sharing all of this across the enterprise, the entire organization can better respond to customers’ needs. “The data warehouse is always desirable, above and beyond a data mart, because it is application-neutral, contains more data, and has more uses than does a data mart,” asserts Dan Graham, an executive with IBM Global Business Intelligence Solutions, located in Somers, NY. “It’s more than a tool for improving customer response – it’s also about optimizing product selection, timing and other aspects of operations.”
However, the scope of this type of project can generate a number of problems. First, even its strongest proponents concede that the cost of deploying a full-blown data warehouse is daunting. For the same reasons, such a project requires a relatively long deployment time – it’s not something that can be rolled out in a few months, nor will the return on investment be swift.
For end users, the scope of the warehouse can translate into poor performance, measured either as the ability to access the warehouse or the speed with which queries are processed. Another issue related to the scale of many data warehouses is the need to synthesize heterogeneous data streams so they can be uniformly processed and analyzed. While it’s one thing to standardize naming conventions, it’s quite another to deal with data that comprises multiple variables that have been calculated using different metrics.
“Unfortunately, many data warehouses deal with `reduced’ statistics and table sets that are amenable to analysis,” observes Bill Ray, president of Group InfoTech, an East Lansing, MI-based IT consulting firm with extensive experience in data mining for one-to-one marketing applications.
Finally, the combination of legacy systems and large volumes of data can create a bottleneck in terms of the number and variety of queries the system can process. According to Ray, the lack of processing power compromised the design of many enterprise resource planning systems. While such limitations can be accounted for in single queries, it causes problems when derivative analysis is involved. “If one command triggers a single report, that’s fine,” he observes, “but if the outcome is used to generate sequential reports, what is the value of the data?”
Not surprisingly, the response of many IT and marketing executives who witnessed early attempts at building data warehouses was to scale back and implement department-level data marts. With this type of project, everything is more manageable. The deployment cost is only a fraction of what is required for a data warehouse, and a data mart application can be deployed in a matter of months. “Data warehouses caused concern that the project would take too long, so why not implement a data mart,” maintains Oracle’s Menninger. He adds that the ability to run marts on existing systems enables business units to deploy them without having to ask corporate headquarters for additional resources.
And since departments are managing projects to satisfy their own needs, the data marts can be built to answer very specific questions on well-qualified data. Notes IBM’s Graham, “You can build a data mart for the Web, one for direct mail and one for sales. Each will be carefully focused and solve the specific problem very effectively.”
However, there are several drawbacks to this that directly impact the goals of a one-to-one enterprise. First is that the knowledge gained from individual data marts has a tendency to remain in that department – even when other departments or divisions could benefit from the same information. “In the long run,” contends Graham, “you’ve built a point solution that results in islands of information.”
Further, one of the problems of the data mart approach is that the quality of information derived has a relatively short lifespan. “Data marts have a half life,” Menninger points out. “Initially, the information they provide is very valuable, but the value diminishes quickly over time.”
Consistent data across the board
Having observed the problems of building massive data warehouses as well as the limitations of implementing data marts, the current trend among marketers and IT managers is to deploy a data warehouse that can be accessed by different departments, which then use data mart applications for their particular needs. This approach, sometimes called a federated or hub-and-spoke architecture, satisfies the need for enterprise-wide access to consistent data, while also providing the operating units with the ability to leverage marts for particular applications.
First, computer technologies have improved significantly, eliminating many of the processing bottlenecks and limitations that plagued first-generation data warehouses. Hindsight is helping, too, as managers overseeing more recent implementations have tried to avoid the pitfalls by aligning the business objectives with the construction of the data warehouses, and by taking an iterative approach to construction.
At Safeway, Plc., a British grocery store chain based in London, the firm made a strategic decision to build a data warehouse accessible by all of its department managers, notes Jeremy Wyman, business solutions manager. Next it assigned IT specialists to work with each department to make sure the system would satisfy its needs. “Marketing people have difficulty asking questions that can be analyzed with data mining techniques,” Wyman notes. “Getting people with the analytical skills to work with the business people to determine what is possible is a critical success factor.”
It’s also important to keep the scope of the project from growing too large, too fast. At the Prudential Insurance Co. of America, a Newark, NJ-based firm that provides a variety of products and services in the areas of property and casualty insurance, banking, real estate, securities, and mutual funds, this approach has proven invaluable. With 30 million customers, 10 million households, 14 million individual policy holders and 35 million active insurance policies, building a data warehouse was no mean feat, and Prudential has been working on the project since the early 1990s.
However, explains Alan Satterlee, manager of Prudential’s marketing performance and information group, there’s no reason the warehouse has to satisfy every application from the outset. “Releasing a data warehouse that can answer a large percentage of questions is the key [to success],” he points out.
With a data warehouse in place, marketing departments can take advantage of the warehouse to perform a variety of analyses, such as cross-selling opportunities between different groups. “You want to do the analysis with the mart and then project the results back into the warehouse,” reports Richard Van Schoick, vice president of marketing services at Prudential Insurance and Financial Services, who serves as the liaison between the marketing and IT departments.
One of the issues that has been brought to the forefront by the Web and e-commerce is that the rate of growth of data warehouses has skyrocketed. Obviously, this places a premium on building a warehouse that is scalable, as well as on data mining tools that don’t easily bog down. As Graham notes, “Multiple customer touch points are the back side of one-to-one that exacerbates the issue for IT departments.”
According to Ellen Joyner, program manager for customer relationship management at SAS Institute, Cary, NC, this is where the role of metadata is vital. Not only does it provide the means by which to ensure consistency among all the entries, metadata also enables other users to learn about prior analyses to avoid performing redundant ones. Further, she argues that the metadata provides the context that quantifies and qualifies such attributes as where the data came from, its age and the types of rules that were applied when it was consolidated into the data warehouse.
Clearly, building a system that can store and analyze massive amounts of information can pay off in the one-to-one arena.
At Safeway, for example, the company tracks all births in the United Kingdom, hoping to cash in on the sale of diapers and other products for kids. However, rather than focus only on mothers, Safeway’s analysis has shown that it’s often other family members who should be targeted with promotions such as direct mail. “We found that it’s the husbands and grandparents who often buy nappies because the mother is dealing with the newborn,” Wyman relates.
Similarly, analysis of video sales has led to a whole new project. Whereas grandparents have traditionally bought clothing for grandchildren, Safeway’s data mining indicated they were increasingly purchasing videos. As a result, the company is now embarking on a joint venture with Walt Disney Co. to market videos to grandparents through direct mail.
A centralized data warehouse that feeds distributed data marts has also proven successful with adults, too. Several years ago, Harrah’s Entertainment Inc., a casino gaming company based in Memphis, TN, stored customer information on separate systems at each of its then-16 casinos in the United States. While this set-up enabled each casino to develop one-to-one marketing applications for its customers, they were out of luck if they visited another site. In response, the company implemented a single Patron Database data warehouse that provides every site with complete information about all customers, so they can be offered the same discounts and benefits in order to increase loyalty. In the year following the roll-out, Harrah’s enjoyed a 73% increase in the number of customers visiting its casinos located in different parts of the country.
Thus, while the tools and techniques have changed dramatically since the gold rush, discovering the mother lode remains as valuable as ever.
It took about two years for 3Com Corp. to consolidate a hodgepodge of 50 databases into a centralized one for managing its customer and vendor relationships worldwide. But in doing so, it fulfilled one of the requisites for being a true one-to-one marketing organization.
The Santa Clara, CA-based manufacturer of voice, data and video communications equipment was unable to identify or differentiate its customer base systematically. Old “legacy” databases at the firm lacked the standardization capabilities necessary for efficient marketing analysis, let alone one-to-one marketing.
“We had to re-engineer the whole business process to support ongoing management of data standards worldwide,” says Jane Lafontaine, global database marketing manager at 3Com.
A worldwide database subsequently was created to keep 3Com competitive and improve its ability to target and communicate with customers individually based on their purchasing history. This data standardization has allowed 3Com to maintain a consistent set of product questions and multiple-choice survey answers for all corporate divisions throughout the world to share when collecting and organizing data. As simple as standardization seems, the benefits were great.
The new database – accessible to about 250 people at different locations all over the world – allows 3Com to build customer records using multiple data elements obtained from e-mail, direct mail and teleservices campaigns. “We get daily feeds into our database,” says Lafontaine.
Each customer record contains about eight data elements linked to other relational database records. The data segments and identifies customer relationships based on such factors as department, division or location.
3Com today is expanding a huge database with more than 7 million customer records. The entire database program was outsourced to Gnomix LCC, in Pleasanton, CA, and Acxiom Corp., in Conway, AR.
The Direct Touch program, as it is called, is based on tracking customer data for multiple business units and product lines. It helps 3Com identify new marketing opportunities and use data for cross-selling.
The new database supports a recently introduced e-mail marketing program, which already comprises about 50% of the total advertising and promotions 3Com uses to communicate with customers.
Among the most favorable results thus far at is the response to e-mail offering 100,000 customers a free software upgrade, available on the Web for customers to download themselves. It pulled a whopping 62% response rate.
And by merging all the databases, 3Com is able to avoid sending duplicate e-mails to some customers, while at the same time unwittingly failing to communicate with others.
To avoid over-promotion, Lafontaine says all divisions with database access collectively send no more than one e-mail per week to customers – and even then only to customers who have given their permission to receive it.
Prior to developing the new Direct Touch program, precious little data was collected by 3Com. The tiny bit that was collected originated from product registration cards, lead generation campaigns and various corporate divisions.
This non-standardized data was of little use for defining markets or interacting with customers. While 3Com boasted $6 billion in annual sales and 300,000 customer records, it had virtually no database segmentation capability.
Customer data at 3Com was in effect inaccessible, as it was scattered worldwide among 100 different spreadsheets, project vendors, multiple business units and various other employee applications. Subtle differences and variations in the way divisions collected and organized data made it difficult for 3Com to develop and maintain a global or consistent view of customers. For example, under the old system virtually no two of 3Com’s product registration cards were organized alike. While one division’s product registration card might ask the number of employees at a company overall, another might ask for the number of employees at a single work site, says Lafontaine.
Network
