Data Mining in the 21st Century
With all the advancements in data mining, why do we keep hearing that direct marketers are experiencing declining results? Direct marketing service companies—specifically, the list industry—are at a turning point: Deliver the performance boosts that clients are demanding by leveraging both the data and data mining tools, beginning with the fundamentals, or someone else will.
A Different Mindset
While many direct marketers who deploy historical response list marketing tactics are experiencing diminishing performance, those firms leveraging advanced data and data mining tools are meeting with continuous improvement in results.
How can marketers get out of the former group and into the latter? First, they and their suppliers need to move from broad terms to the specifics of data analytics, while aligning practices and results. Over-use of the term data mining—like that of the term CRM—has caused it to mean all things to all people. Hence, it’s referred to as a multibillion dollar opportunity filled with experiences of non-quantified failures, or declining results. For this discussion, let’s put data mining in the context of exploration to boost performance, or more specifically, to increase marketing ROI.
In addition, when you hear direct marketers are experiencing declining results, this statement must be put into context. Within the direct marketing industry, list rental revenues have climbed into the billions of dollars, which in turn has driven hundreds of billions of dollars in product sales. This is where the finger is being pointed as the reason for such dismal results. Consider that the majority of list revenues follow the supply chain model from list owner to list manager, then to the list broker, on to the product seller and finally to the buyer. Minimal data mining occurs in this supply chain model.
As a result, direct marketers continue to challenge the list industry to find higher volumes of better-performing lists. The reaction to this challenge has been to add more lists—which are assumed to have certain attributes—into the merge process. This is where the finger accurately can be pointed as the cause for declining results. But the technology of merging many small response lists at 50 percent to 80 percent de-dupe rates, (containing prospects only reported to have the attributes of a potential buyer) is not data mining.
Data mining in acquisition marketing is the identification of attributes across an experiential set of solicitations, responders and converters that discriminate between those prospects who have exhibited the potential to respond or convert and those who have not. This type of data mining, done correctly, yields predictable results.
A data mining paradigm has occurred with an exponential increase in “atomic” (i.e., single element) or transaction-level data; the tools and experience already exist to leverage this new, powerful information to provide marketers with better performance across greater volumes of prospects.
The most immediate aspect of data mining for direct marketers is to begin making better use of the available transaction and behavioral data, turning compiled data into response lists. The evidence of quantitative boosts in campaign performance (increased prospect volumes and the dramatic reduction in archaic merge processes) is abundant. Let’s review several examples across a range of goods and services.
Bar None, a national automotive credit service organization that uses broad market media, the Internet and direct mail, has been driving buyers into auto dealerships at an amazing rate. Deploying credit-based models, Bar None is achieving response rates greater than 1 percent. This is double the performance rate other marketers are realizing in marketing to prospects (excluding former dealer customers) for dealership clients. Where much of the automotive industry targets by distance to the dealership, Bar None created a more precise prospecting approach in targeting by driving distance, prior auto ownership history and a set of credit bureau-specific models that predict a buyer’s likelihood to respond to a direct channel offer as well as to convert.
According to Dan Staub, marketing executive for Bar None, “We are just beginning to tap the power of this high-level targeting ability. We expect that advancements in personal messaging, multichannel [e-mail and direct mail] campaigns and continuous model refinements have the capability to further boost performance across larger prospect bases for our dealer clients.”
Scotts LawnService, an on-site lawn, tree and shrub care service—and an organization that already understands the marketing power of data mining—was intrigued by the ability to integrate demographics with property data (e.g., lot size, purchase price, square footage). Using a base of recent responders and non-responders, a model was created from hundreds of atomic demographic and property elements to identify prospects most likely to respond to a direct mail offer for lawncare service.
The challenge was to exceed current model performance by at least 25 percent. According to Scott Jablonski, marketing manager at Scotts, “We are testing new data sources and a new type of thinking. We’ve just begun to tabulate the results from our spring campaigns and are anxious to see whether results meet their model forecasts.”
Robert Groessner, CEO of direct mortgage lender HomeStar Direct, is a disciplined direct marketer looking to boost his firm’s campaign performance to the next level. A structured test of high-performance data—more than 300 individual-level credit elements modeled to identify interactions between variables, e.g., home equity, revolving debt, age—was launched, and the results were 30 percent to 40 percent higher than head-to-head testing against current control programs using creative and copy approaches developed based on more traditional models.
Groessner says, “Testing [new data approaches] not only convinced us to proceed further, but it helped us realize that the reason for this success is the combination of disciplined database marketing and ‘out-of-the-box thinking’ that we’re able to do with such an in-depth level of available data.”
Good Marketing Is Continual Innovation
This brings us to the next most important aspect of data mining that direct marketers face. It has to do with a question that one of my company’s most advanced clients asks me each time we meet: “What are you going to do for me next?”
The “next” will vary for each marketer, depending on its stage of marketing sophistication. But what is clear is that marketers must step away from the age-old practice of merging multiple response lists to develop prospecting files. The technology is here to do more, and those practicing disciplined analytics are seeing quantifiable, continuous performance improvements. So, what are you going to do next?
Peter Harvey is founder and CEO of Intellidyn Corp., an analytics and database marketing firm in Hingham, Mass. Harvey is a former Fortune 500 marketer with successful direct marketing campaigns for firms such as Bank One, Chase Retail, and NYNEX Corp. He can be reached at (781) 741-5503, ext. 100.