Double Take
Want to clone your best customers? First take a closer look at what you see in them.

More of a good thing. It’s why many companies target acquisition efforts based on the attributes of existing customers. If these people buy from us, let’s just go find more people like them.

It’s a simple idea, and it’s often executed in equally simple ways — such as a “straight-select” of Glamour’s subscription list because your target customer is a 32-year-old female earning $50,000 a year. But this is a little like hiring a new assistant based solely on her resume.

If you really want to genetically engineer new best customers, data experts suggest a more scientific approach — a customer acquisition model. In data-speak, it’s a mathematical representation of characteristics that define a given subset of your customers. In English, it’s a way to identify prime prospects with precision vs. presumption.

Why do I need a model?
“A model helps you avoid making assumptions about the characteristics that define your best prospects,” says Bill Schneider, CCG senior database analyst. “It allows for things you may not know about your best customer — or combinations of attributes you may not think about.”

Yes, it will cost more to build a model. But you don’t have to dive in at the deep end. “Typically, you’ll just commit to purchasing a certain number of prospect names to test,” Schneider explains. In the process, you’ll get a wealth of insight about your best customers.

With CCG’s New Customer Acquisition Model, “You’ll get a profile of characteristics like income, age, homeownership and number of kids,” says Julie Pfeifle, CCG senior database analyst. “You’ll be able to see how your customers compare to the general population.” Since you typically don’t have the prospects’ transactional or response histories, acquisition models use lifestyle, demographic or lifestage data to bridge the gap from your best customer to look-alike prospects. So you may in fact find that your Glamour girl is a fitness buff or a foreign travel fanatic — revelations that can make a big difference in how you attract her business.

With CCG’s model, you can even zoom in on attributes of the “best of the best” — the top-scoring names from the prospect list.

How does it know who to clone?
Whatever customer behavior you want prospects to emulate, the model helps you identify the people who are most likely to act that way. Defining that model behavior “can be based on recency/frequency/monetary (RFM) data,” says Schneider. “But it also can be based on customer profitability or be the people in the top tier of a loyalty program.”

Your clone-worthy customer may even be based on a certain promotional objective. “If you’re a catalog retailer and you’re doing a holiday promotion, you might want to find look-alikes for customers who bought last holiday season,” says Pfeifle.

For one online stock brokerage, “best customer” meant someone who not only opened an account but also executed trades and brought in large balances. “This client had been using vertical lists (such as magazine subscription lists) as its prospecting pool,” says Greg Sultan, CCG VP/national sales. “When they came under pressure to increase the number of accounts and decrease the cost per account, they began modeling based on the propensity of a prospect to respond and open an account.” The results: Cost per account decreased 40 percent, while mailing volume increased 330 percent.

Is this an exact science?
The most common mistake in best-customer cloning is trying to outthink the model. “You can create self-fulfilling prophecies,” says Sultan.

“You might think someone with a high credit score won’t respond to a certain promotion. But don’t pull them out of the model until you’ve proven that they won’t respond. Let the model figure that out.”

Schneider agrees. “The thing about modeling is that it’s kind of a black box — you don’t really know what’s going on in there,” he says. “You and your vendor need to keep saying to each other, ‘Here’s what we’re seeing; does that make sense to you?’”

Other cloning caveats:

Be careful what you wish for. “You don’t want to model the wrong behavior,” Sultan points out. One retailer created an acquisition model based on RFM data, mailed to the prospect list that best matched this model and saw sales go through the roof — then drop again three weeks later. The culprit? Returns. The company had neglected to take high returners out of the model up front.

Other variables can skew results.
“If you offer aluminum siding with a home equity line,” notes Sultan, “that may not appeal to a guy who just painted his house.” So if at first you don’t succeed, don’t automatically throw out the model. “Your model may be accurately ranking people,” Sultan adds, “but your creative or your offer may be affecting the results.”

What you see is what you get. “If your model is built on results from a past direct mail campaign,” says Sultan, “it will probably only predict for that same communication channel.” So factor in the customer’s preferred means of responding.

Resist the temptation to generalize.
“Some companies try to clone best customers as a homogeneous group,” says Sultan, “but there may be important differences within that group.” For example, one regional financial services provider found that its relationship-account holders included both seniors who preferred banking in person and a younger set who liked banking electronically.

Is acquisition modeling an exact science? No. But the more accurate and relevant the information you feed it, the more scientifically valid your model will be — and the better chance you’ll have of cloning your true best customers.

For more information about any of CCG’s data analysis tools, contact Senior Vice President/Strategic Consulting Lane Ware at 1.800.525.0313 x106 or send an email to lane@customer.com.