What does Rumpelstiltskin have to do with database marketing? An early chapter in Spinning Straw into Gold: The Magic of Turning Data into Money, a new e-book by John Miglautsch, makes the connection clear. τ The chairman of Miglautsch Marketing in Wisconsin, Miglautsch has built marketing databases for companies such as Moore Business Systems, Hudson’s Bay, Cabela’s, Baseball Express, Adobe, and eTrade. His book (available at http://www.migmar.com) offers insights into database marketing that are, indeed, worth their weight in gold.
Spinning Straw into Gold starts with the premise that database marketing is often a zero-sum game: “Turning data into dollars is the exception rather than the rule.” This is especially true if you rely too heavily on number crunching and too little on relating to the marketplace.
In discussing the fairy tale “Rumpelstiltskin,” Miglautsch observes that “there are two tests and a rollout in this story” (conveniently reprinted in full in his text). Moreover, “Rumpelstiltskin” is based on “two big lies,” the first that straw can be spun into gold, the second that the miller’s daughter could do it.
In database marketing, there are also two big lies. The first is “We know who you are” based on rich data. Yet response data are infinitely more valuable than compiled data. Since most database professionals know this, they don’t usually buy this first big lie.
The second big lie is “We know what you’ll buy.” The chemistry of making buying decisions is unpredictable. People intend to buy one thing in a store and end up buying something else, or unrelated things on impulse. If we can’t even predict what we ourselves are going to buy, how can we pretend that we can predict what other people are going to buy? As Miglautsch says in the book, “Think hard about that trip to the store the next time you listen to a slick presentation on the virtues of neural-net relational client-server massively parallel star-clustered genetic-evolutionary algorithmic modeling.”
Most of your data are worthless
The “straw” in the Rumpelstiltskin story is like data. But recall that straw is not hay. Hay goes into an animal as food (it is dried grass), while straw, which is the useless stalk of a wheat plant, is “what goes under an animal and catches what comes out of it. You have to understand that like straw, most of your data is worthless…If you can face that, you are far ahead of your competition in turning data into money,” writes Miglautsch.
Data are worthless in part because they are incomplete and inaccurate. Order-entry systems are not database marketing systems, and the unreliable and inconsistent data that come out of virtually all of them are rarely a good foundation on which to build a marketing database.
To make any sense of database marketing, Miglautsch argues, requires meaningful variables: “The single biggest reason why people are not turning data into money is that they are not building enough variables.” A variable is a simple abstraction, a “summarization” of data. IT professionals define it as a column in the database, but it is much more effective to see it as something derived or calculated from the data in the database. Once you have more than a half-dozen variables, however, the only way you can usefully evaluate them is by modeling, which brings variables together in a meaningful structure.
But, says Miglautsch, creating variables requires thought, which is work, and that “discourages most people.” Even beyond thought, though, what you really need is insight.
A little modeling magic
Spinning straw into gold requires “magic”; turning data into money requires the magic of exceptional marketing intuition. A prospector might discover gold lying on the ground, and a scientist can try to figure out where to dig for it. But only a marketer can spin data into gold by using it to leap in the right direction into the future. According to Miglautsch, “Accountants tabulate the past, while marketers use data to try to see or create the future. Models help them see the data as clearly as possible in order to do that bit of magic.”
Of course, magic in database marketing involves modeling, in which Miglautsch identifies three levels. Level one entails finding prospects just like your best customers — in other words, profiling.
Level two is a matter of mailing only to the most profitable customers — also known as “profit improvement.” Unfortunately, profit improvement too often comes at the expense of growth. Modeling helps you get beyond the “flat” response curve that recency/frequency/monetary (RFM) analysis yields, but most modeling is built around some form of linear regression. Within whatever linear ranking (or variables) you create, there will always be a range of customer behavior.
To find the most responsive and most valuable customers among multiple variables you need to use a tool like chi-squared interactive detection (CHAID) analysis. A CHAID model created by Miglautsch Marketing for an outdoor-merchandise catalog, which management agreed accurately profiled who its customers were, generated $2.4 million in extra profit. Says Miglautsch, “If I were to boil down spinning straw into gold, why we have been able to make money with data, it would be because we were able to work closely with people who understand what the customers are like.”
If level-one modeling identifies the “I love you” factor, and level-two eliminates the less-responsive customers in each variable in the profile, then level-three modeling is a way to supercharge your promotion strategy to create more good customers.
Here’s the dilemma: Every frequent, high-value customer started off as a one-time buyer, and maybe a low-value first-time buyer at that. You can’t focus on your high-end cells at the expense of the low-end cells, nor can you do the opposite, by trying too hard to promote low-end cells to high ones. And you certainly can’t reposition your catalog or your offers too far from the formula for success that has been working all along to “promote” customers up the RFM and lifetime value (LTV) ladder.
Good customers “connect” with your offer. Customers say, “These guys speak my language” or “I like their products” or prices or whatever it is that generates the “I love you” factor. But most of your customers are not “in love” with you; they are just customers. So, the book says, “The key to spinning straw into gold is understanding the tension between keeping the good feeling of your good customers and changing something to connect better with at least some of your bad” or not-so-good customers. But “you do have to change something” for response rates to change.
Doing something may mean identifying and responding to new currents in customer priorities and needs, which sounds like data mining but is not. A data-mining platform can often fail to provide a way to pull names for mailing, and “the most minimal system requirement for spinning straw into gold is that the database must flawlessly pull names.” Changing the metaphor slightly, Miglautsch suggests that we are not “prospecting” for gold in the mountain but rather “sluicing” for it, like panning in a stream. “We generally know what we are looking for, we know the programming process required to find it, and we know when we’ve found it. This is a far different process than data mining,” where the object of your search is usually unknown at the outset.
The “hard work” comes in when results from data analysis must be translated into offers that make sense. The better your systems, the better your data; the better your data, the better your analysis; and the better your analysis, the better your offers can be.
In the end, the “magic” is in the marketing, in coming up with a new twist to keep your product mix fresh and your offers appealing without abandoning your marketing personality — encouraging customers with low RFM/LTV up the value chain while not forsaking those who are already at the top. It’s a balancing act, helped along with the ballast and balance of good data and focused analysis.
Spinning Straw into Gold is obviously not a how-to manual in the usual sense. But it is a primer in the very best sense, meaning that if you don’t understand and appreciate what John has to say in this book, there is very little chance you will ever be successful in the practice of database marketing.
Ernie Schell is the president of Marketing Systems Analysis, a data technology consultancy based in Southampton, PA.