The list compiling game isn't as easy as it used to be. Once upon a time, yellow pages and public records served as almost one-stop-shopping destinations for those looking to build files. They had to verify everything by phone, but the information was there.
Today, thanks to mobile phones and VoIP, many new households simply aren't popping up in telephone listings. And because of increasing restrictions, access to motor vehicle and other public data simply isn't an option.
So what's a poor data professional to do? Easy. They get creative, offering more options like modeling, segmentation and other value adds to make themselves more of a partner in their clients' marketing programs. Direct recently gathered several compiling industry pros together for a roundtable in our New York offices. Our Richard H. Levey moderated the discussion, which covered many topics, ranging from the convergence of business and consumer information to how to cater to clients' increasing demands.
‘If you're just in the business of ‘name-address-phone’ as a commodity, you're going to be in trouble,’ according to Experian's Ed Connolly. ‘You have to add value to your data elements.’
DIRECT: What do clients want from compilers?
ROVELSTAD: We're seeing an expansion in the multichannel arena. Retailers are looking to contact their customers in the mail, through the stores, by phone and online.
KHANNA: Multichannel is the growth area. Most of our customers are seeing that the response rate is so much higher than it is with just one channel.
BOWMAN: We're seeing a lot of demand for granular and industry-specific data — not just behavior but attitudes and hand-raising. And this data is informing larger decisions. In retail, it influences inventory, real estate and merchandising.
KOWALSKI: Our customers are asking us to do more with the data. They're not just asking for a list, they want us to help them with a whole year's worth of information, to build models or help them run an entire campaign.
BOWMAN: Historical data is another thing we're seeing a lot of interest in. We're going back in time and [looking at] trending information, trying to understand how a particular behavior four years ago can be replicated.
CONNOLLY: We tried to do work with historical data in the early 1990s, but clients didn't have enough history on their customer bases. Now they do and they're interested in pursuing lifetime value and trending. They're also tying in transaction data and looking at softer measures of consumer behavior — like brand loyalty.
GREENBAUM: We specialize in the life-cycle market, and we're watching to see what a consumer does after a lifestyle change. We see segments that in six months are doing this, and in a year they're doing that. We look at the best time to focus on a new homeowner or new mover.
CHANGING NEEDS
DIRECT: Are you pulling away from any products or services?
KHANNA: We've seen a considerable drop in people just using data for telemarketing purposes, obviously because of the do-not-call list.
BOWMAN: We've seen a pullback from date of birth and other sensitive data attributes in certain industries. Today, you might want to put that type of information in a model and use it in a little more discreet way.
GREENBAUM: We have a lot of clients that would do mailings and would follow up on the phone. They can no longer do that.
CONNOLLY: Unless you have an existing relationship, it's pretty tough to telemarket. Telemarketers are used to having a 6%, 8% or 10% disconnect rate, and the rate is considerably higher if you only call those available after the do-not-call list.
ROVELSTAD: Some are expanding online.
KOWALSKI: We see that as part of an overall marketing program, but we're not seeing a flood of online prospecting. E-mail is being used as a follow-up tool, not as a prospecting tool.
PRIVACY
DIRECT: Speaking of privacy, how are you managing that issue?
KOWALSKI: The challenge is making sure that consumers aren't getting scared about the information we have on them, and showing that we're adding some type of value and that we're not just white noise. You need to respond quickly and honestly to inquiries and complaints, follow DMA guidelines and make sure your database has all the proper firewalls and restrictions.
KHANNA: We've all done a pretty good job of having a privacy policy in place, along with a process to monitor every mail piece or phone script, especially when the data set is a sensitive segment. You tell the marketer that you don't want them to mention somebody's age or that they have a child. We want to help them be a little more educated and have a higher response.
BOWMAN: [Checking your clients' credentials] is a big thing on the front end. You want to do business with people who will guarantee good policies and practices.
KOWALSKI: Ed mentioned healthcare and ailment data, and they're almost as sensitive as Social Security numbers. You want to make sure that if you're using that type of data it isn't popping up on an outer envelope and that you're complying with any [regulations] specific to that vertical market.
CONNOLLY: You also have to identify what data elements your company feels are sensitive. Ailments are one, children's data another. We don't allow Social Security data out at all on the marketing side, and if the client does send it, we encrypt it. But you also have to look at things like age. If a mailer is targeting people age 65 and over for a scam, you want to put a process in so that somebody automatically reviews the mail piece and offer.
DATA SOURCES
DIRECT: Motor vehicle information hasn't been available since the Shelby Amendment. What impact has that had?
BOWMAN: You just have to find more sources. Instead of just having one easy source of that information, you have to go out and find self-reported data in other channels.
KOWALSKI: Data from one source is not necessarily going to be reliable. In the past, you could count on getting that information from one place — like public records — where somebody would tend to put their correct date of birth. But if you're now getting it from the Internet or self- reported means, you have to verify those sources against each other.
ROVELSTAD: They're also using predictive modeling when they can't get self-reported information.
KHANNA: It's strategic. We've done a lot of acquisitions in the brokerage and management arena, so we can go beyond compiled data. In 2001, we were aggregating 800 million transactions to compile our database. Today, we've done more than 3 billion in just a few years. The historical delineation between compiled data or response data has become fuzzy.
GREENBAUM: We're focused on life-stage data, so it's not as relevant to our compilation process whether the data is, say, from a subscriber file or a mail order buyer file.
CONNOLLY: When age data is self-reported, it might be accurate plus or minus five years. Age isn't nearly as accurate as year of birth.
DIRECT: What's happening with compilation costs?
KOWALSKI: You have to look at each data source. Many sources specialize in a group of people of a particular age, so you have to go to more sources to compile a complete database.
BOWMAN: You have to engineer the data more intelligently to bring in more sources. It does increase the cost, but at the same time it increases the value so the price doesn't get compressed. If you let it be viewed as a commodity, it will be. Perception is reality.
GREENBAUM: We're finding that it's more expensive to compile when the market is not allowing incremental price increases. It's very competitive. There's a lot of commoditizing of data being done.
DIRECT: Lynn, you're a reseller. What's your perspective?
ROVELSTAD: We enjoy a relationship with just about everyone around this table. The compilers have gotten more efficient, and technology advances have helped to keep some of those costs in place. But we have seen the cost come down as more compete.
KHANNA: The solution is different for different segments. For the mom-and-pops, it's about providing a complete solution with data and helping them execute a campaign. We're trying to stop the commoditization of the base ‘name-address-phone.’ It's very easy to ship [telephone] books to another country and have ‘name-address-phone’ keyed in. The question is: How accurate is that? If you don't have those fundamentals right, you'll never match up the sources.
CONNOLLY: We're seeing the same thing. Customers are asking for more, and they're trying to compress the price.
DIRECT: Are they doing so successfully?
CONNOLLY: If you're just in the business of ‘name-address-phone’ as a commodity, you're going to be in trouble. You have to add value to your data elements.
BOWMAN: The days of just delivering data are gone.
GREENBAUM: And when you add value, you don't always get to add price, so it's hard to keep up sometimes.
ROVELSTAD: But at least you don't lose ground when you add value.
GREENBAUM: Right — that's why you have to do it. Some people use the suppliers that are commoditizing and then they say, ‘I tried compiled data and it didn't work.’
KHANNA: People often fail when they make their decision based on price. Maybe it's because the name and address weren't right. But they get totally disenchanted with direct marketing. If the first campaign isn't successful, the typical response we see is that the marketer won't even try a second one for 18 months. How many small businesses stay alive that long?
GREENBAUM: And that just doesn't happen to the small mom-and-pops. Some larger mailers think that compiled files don't work. I say, ‘You need to look at this as a source of data because your response data pool is shrinking.’
BOWMAN: We have a transition process for clients to go from using vertical to compiled data. There's a testing environment where they can evaluate each list they're buying. It may not get rid of all their vertical lists but they will save enough money to see the benefit [of using compiled data].
KOWALSKI: On the other side, you have large customers that know direct marketing has to be part of their mix. They've been hammering away at it for so long that they've just come to accept these horribly low response rates. You have to come back and show them that you can understand their customer base and [give them] some tools they may not have used before. Sometimes they'll see a 1% or 2% jump in response, [which can be impressive] for the volumes they're doing.
DIRECT: It sounds like there's been a substantial move away from RFM into softer areas like psychographics.
ROVELSTAD: It's a combination.
KHANNA: RFM is still there, but we're seeing four trends: lifestyle, behavior, life stage and multichannel.
BOWMAN: The information technology is such that more data can be matched together, and that's pushing a lot of these demands.
ROVELSTAD: Modeling is now available to the small mom-and-pop shops as well as the big guys.
KOWALSKI: You can turn around a model in a couple of days now, whereas in the past it used to take weeks or more to build a decent one off a customer file.
KHANNA: Even beyond that, there's enough actual behavioral lifestyle information available that in some cases you don't even have to model it.
DIRECT: From which industries are you seeing the most interest from mom-and-pops?
KHANNA: There's B-to-B. A lot of small businesses provide services to other businesses.
CONNOLLY: We've found it beneficial to identify home business owners on the consumer side. Linking those together gives you firmograpahics and demographics about that small owner.
DIRECT: You've mentioned that you're getting more granular. Yet most of you offer some sort of clustering analysis. Are both still equally important?
CONNOLLY: They actually go together. Sure, you want the more granular level — transaction data — that will allow you to tell what people are doing. Then you can use that to generate models. And granular data can help when you're dealing with a client file. You can use it to model data parts that don't match.
KHANNA: Part of it is also transitioning from a total clustering mentality to being able to use individual lifestyle indicators. It will take time, because historically there hasn't been good coverage. Previously, if you wanted to target golfers, you'd be going after a particular cluster. But a 30-year-old golfer with a certain income is a different buyer from a 60-year-old golfer with the same income.
BOWMAN: Clustering systems are a beginning framework for managing your view of a customer relationship. But you need to get granular to execute. This is something that's difficult to explain to marketers. When you use a clustering nickname with associated behaviors, it provides a common language for talking about a profile of a particular segment. But you've got to get other data to slice those clusters a little bit thinner.
KOWALSKI: It used to be that we could just take and label everybody over 65 as a senior. But now you've got clusters within that group. For example, you can't just call somebody a boomer now and know what their attributes are going to be. Are they an aging boomer, are they an emerging boomer? Are they a boomer who has extra money to spend? You've got to take all that stuff and make clusters within that.
BOWMAN: A best practice with clustering systems is to use them in conjunction with customer models. Off-the-shelf segmentation can be overlaid to dig deeper into low-performing deciles.
DIRECT: What are the challenges in business-to-business compiling?
KHANNA: Historically, you could compile B-to-B information from the yellow pages and then verify it by phone every year. We've used a multisourced approach for five years to prepare for shrinkage in yellow pages coverage. To give you some sense of metrics, a few years ago we noticed a slight drop in our physicians' coverage. It fell to about slightly over 500,000, where historically it had been 650,000. Yellow pages advertisers are realizing that if they're going to be fiftieth in a category, maybe they don't want to advertise there. So we've added a lot more sources to our business file and our phone-calling process now is really for both verification and research. Five years ago, we'd barely do about 15 million phone calls annually to verify the database; last year we made 25 million calls. To some extent it's becoming like consumer data.
ROVELSTAD: We're seeing more people applying techniques they've used on the consumer side to the business side. They're creating business-like clusters as well as using traditional modeling within the set of available selection items.
KOWALSKI: Our B-to-B customers are using a push/pull strategy in the healthcare market. They're not only marketing on the consumer side but to physicians as well.
CONNOLLY: Much of our focus is on smaller businesses. We apply analytics to business files too, and that's going to happen more and more. You'll need to have more sources and verification. And you'll have to do more segmentation, modeling and firmographics.
BOWMAN: What we're seeing from clients is a blurring of the consumer and business worlds, linking consumers to where they work and vice versa.
KHANNA: The Internet is an interesting way for businesses to give us information. They can say, ‘Hey, I'm not being listed on many of the search engines. Can you make sure my business is there?’ They give us that information, and we'll research it and phone-verify it to complete that record and provide it to our partners. It's becoming an important way for us to validate our information and get the coverage. Small businesses want to be sure they're included in directories because that's a way for them to get customers. As for firmographics, one key variable is the number of employees. With that you can judge the size of the business.
DIRECT: We've covered almost everything. What else is new and interesting?
KHANNA: Photographs of businesses. We're taking them. If you have a particular destination, a search engine can give you the routing to find it and also show what the area looks like. We're taking both wide-angle photos of the buildings as well as close-ups of the entrances. On the consumer side, it's lifestyle and behaviors. Our strategy is to collect that information through the companies we've acquired in the mailing list brokerage and management space.
GREENBAUM: For my niche market, it's the cell phones and VoIP, and not so much because of the actual phone numbers themselves. New households are springing up with cell phones and [no land lines], so we're not getting an actual address, we're not getting a listing anywhere. Combine that with the loss of the NCOA files, and [it's hard] to tell if people are moving. If someone has a cell-phone family plan with five or six members and you can only identify one member of the household, that makes it a challenge to integrate information about that household into your database.
CONNOLLY: Land lines are dropping at 4% or 5% a year and the do-not-call list restricts who we can call. [It's a problem] because for years the telephone was key to us [being able to maintain] up-to-date or renewable data.
ROVELSTAD: Something we're seeing in its infancy is that quite a few of our printer customers are trying to match trigger information with variable-data print solutions.
KOWALSKI: Customers are asking for transaction data, not so much for the transaction but for the fact that it shows a person is at an address at a certain time. And that's probably a function of public-record information being restricted. Activity at an address can help you verify if a person has or hasn't moved.
BOWMAN: From a compiling perspective, we [need to consider our] relationships with consumers. Much like the credit bureaus have had to become transparent to consumers, giving them access to their own data — and they're actually making a lot of money providing access to the data — I think our industry needs to focus on doing that. We need to understand how we can fortify our relationships with consumers, show them direct marketing's value proposition and make our data assets better.
PARTICIPANTS:LYNN ROVELSTAD
President, AccuData Integrated Marketing
RICHARD H. LEVEY
Senior writer, Direct (moderator)
DEREK BOWMAN
Product leader, Acxiom
AMIT KHANNA
President, database compilation, InfoUSA
CASEY J. KOWALSKI
Director of data acquisition, KnowledgeBase Marketing
JULIE A. GREENBAUM
Vice president for sales, HomeData
