Database Development: Why You Need to Take a Second Look
The more a publisher knows about its customer universe, the better it can strengthen its edit mission, identify new customers, present comprehensive profiles to advertisers, and identify product and marketing opportunities that originate from an internal collection of data rather than an external one. Yet product trends, the most profound being Web-based, are often chased without regard to how customer data captured from one medium relates to data from another. A publisher can find itself with multiple product silos without any operational synergy behind them or an efficient way to access and manipulate customer data to tie together related sales or marketing opportunities.
The solution? A centralized repository of data with a robust architecture that supports and combines all product lines and can be mined for new launches or marketing opportunities. Folio: talked with three publishers about what it takes to build out a customer relationship database;and when it makes sense to create and maintain it yourself or seek the help of a vendor.
"I started statistical acquisition models and the test was so successful we lifted our response rate by 85 percent." Hairong Crigler, VP, Database marketing, TV GUIDE
Where to Begin?
Every publisher has unique database needs. Determining to build one in-house versus outsourcing depends on a number of factors;both strategic and economic. "I think the technology and software have come such a long way that it allows you to focus on your business processes rather than being totally distracted by [assembling] the technology," says Tina Steil, chief technical officer at Meredith Corporation. "I would recommend that you outsource. Only so you can get your house in order, because it's going to take you a year or two to get that going and understand what you need to accomplish without worrying about having all the boxes configured, software installed, maintenance and so forth."
Outsourcing, says Steil, may look expensive at first but keep an eye on long-term ROI. "Short-run, the outsourcing model is going to look more expensive and you're going to think you can do it cheaper internally. Nope. You'll be so distracted with the technology the first year, you'll lose sight of what you're trying to accomplish."
Steil maintains Meredith's database in-house;but it's enormous. Built over many years, it has over 85 million records, some with up to 2,000 different data elements on each individual and an average of 300 per customer. Part of that data depth comes from partnerships with data vendors who overlay their own data with Meredith's, giving a more complete picture of each consumer.
How long before you see a return? That depends, says Steil. "What are you trying to accomplish? Are you looking for a totally integrated solution? Do you have a Web presence you're bringing in? It could be anywhere from three to five years," she says.
Nevertheless, there are some immediate results that, once active, a database can provide to mitigate the initial expenses. Hairong Crigler, vice president, database marketing, for TV Guide says her database of about 3 million unique customers took about two years to build out at a cost of a "few million" but began seeing results within two months after starting in 1999. "I started statistical acquisition models and the test was so successful we lifted our response rate by 85 percent," she says. "While we were building the database I could save the company $1.5 million on the direct mail only."
That trend has continued for Crigler. Last year they sent out 18 million direct mail pieces and saved that same $1.5 million. "The cost of processing is so huge. A data vendor charges us $20 per thousand to score the list. If you mail 18 million, that's a lot of money," she says.
For publishers the size of Meredith and TV Guide, bringing database operations in-house with supplemental help from data vendors makes sense;if only because of sheer volume.
Crigler echoes Steil's recommendation that smaller publishers, who will most likely have basic account information for their customer base already in place, consider outsourcing the expansion of their data. Publishers with, say, 200,000 to 500,000 customers are on the small side and the costs associated with hardware and software acquisition, not to mention the staff required to mine the data and generate statistical and behavioral modeling reports could be prohibitive;and depend case by case;says Crigler. "For smaller publishers it's worth it to have something there, but having a third party maintain it at a minimal cost would be more beneficial."
Maintaining the basics of a smaller data warehouse internally won't cost more than $40,000 to $50,000, says Crigler. However, where the rubber meets the road is in elevating a basic data warehouse to one that connects the dots between customer relationships and behavior. "The issue is whether you have an internal relationship established or not," says Crigler. "If you don't have that then you need help to build that up."
Pricing, says Crigler, is typically based on volume and frequency;volume meaning number of customers and their associated data points, and frequency meaning the number of times you need to access the data and generate reports.
Different vendors have different data. They each collect it in their own way, usually amassing millions of records with extensive demographics along with lifestyle and behavior data. Quality varies and the data points may or may not match up with your customer information needs.
Shop a snapshot of your data around to different vendors;Experian, InfoUSA and Acxiom are three big ones;and have them match your file of names and addresses with theirs to see how they measure up. From there, vendors can append information they've collected to yours, essentially augmenting your customer data in ways that will provide further insight into their behavior and buying power. "And you can utilize that information by either doing segmentation, modeling or a profile analysis to target better," says Crigler.
"If you don't transform it or normalize it you've just got a library with all the books and no card catalog. You have no idea what you're looking at."
Joe Fierro, Director, Information Marketing, VNU Business Publications
Under the Hood It's All About the Data
Joe Fierro, director, information marketing for VNU Business Publications, was hired in 2003 to build a three million-name customer data warehouse. "It's an audience database, so it has all our print publications in it, all our face-to-face conferences and all our online, whether they're e-newsletter or online subscribers. It includes any type of transactional data they might do;if they buy a PDF or they buy a tote bag, we capture that."
VNU's strategy was to have a single view of the customer to understand how they moved among the company's various product platforms. "The problem was," says Fierro, "the information was all over the place and not in any set format. My job was to bring it all together in one centralized database."
According to Fierro, publishers need to keep their eyes on the right ball;their data. "Everyone talks about the return on investment, they talk about the vision, the tools, the platform;they talk about everything but the data. And that's exactly where it starts."
Fierro says he spent 90 percent of his time sourcing VNU's customer data, figuring out where it came from, its format, the fields it belonged in and its value. Then he mapped it. "Is there any core set of data elements that make sense across all the products? Is it useful just for this business or is it useful across the board? How do I make it useful for everybody?"
From there, the data needs to be transformed;or formatted in a way that's of value across the various product platforms. Without this step, without the relational dimension to the data, all you have are names. "If you don't transform it or normalize it you've just got a library with all the books and no card catalog. You have no idea what you're looking at," says Fierro.
Determining data value requires some help from corporate. "You need to know what the vision of senior management is," says Fierro. "And then you have to be very data savvy."
The savviness comes into play when deciding how each data element fits into a larger picture. "Is it just valuable to that particular brand or is it valuable in some other larger context? If it's not, to just throw things in there because it's there gums up your system," he says.
For formality's sake, VNU runs a SQL Server 2000 database on a Windows Server 2003 platform. But Fierro points out that, once again, data matters, and once you have the data structure in place, you've got a platform that any tool can sit on.
The front-end, however, is a "touchy thing," says Fierro. Publishers will likely have different opinions about how the data is accessed by employees and through what interface. Fierro prefers to be the gatekeeper. "If someone doesn't know SQL or any programming languages and you build a front end for them, what you're really doing is building a report because it's all summarized data. And it kind of negates the idea that it's really a database because I've already figured out exactly what you're going to ask."
To Fierro, simply providing a key for, say, the marketing team to click a few buttons to cross one data set with another predetermines data usage in a restrictive way. "I don't see that as using a database. That's simply a report. You're using a database when you're asking any question that you want that spans across all the tables and fields and you learn something new," he says.
For both Meredith and VNU the database has become central to uncovering new business opportunities;in fact, Meredith's $350 million purchase of Gruner + Jahr's Parents, Child, Fitness and Family Circle was critical in filling in younger-customer age gaps in the company's database.
Meredith's Steil says that when corporate is hashing out a new launch or acquisition the database is key in the decision process. "The number-one criteria we put up against [the database] is not reader interest, is not advertising interest, it's if there's the ability to build and maintain a sustainable, profitable circulation model in excess of one million ratebase," says Steil. "Unlike some other magazines that can launch with a newsstand model or with an advertising model, for us it's got to be profitable circulation with a one million ratebase first and that's why this database plays such a huge role in our evaluation process."
"I've found in multitudes of 30,000 targets that we never even went after or thought about," says Fierro. "If you start understanding that you have people getting all kinds of products and services from you, you can go in there with a package for them."
Bottom line is a customer database built on a relational model not only illuminates who your customers are, but their potential for working their way deeper into your product mix. "You can't manage a business with intuition," says Fierro. "You can't build something and then have people say, ï¾‘Well, I always know who my audience is.'"