Cluster Coding


For the past 15 years, Claritas has done a brilliant job of dividing the U.S. population into 62 clusters with catchy names like Shotguns & Pickups and Pools & Patios. These clusters roughly reflect household income and living style. For some products (although not for most), the cluster someone lives in has something to do with the products that person tends to buy. People in some clusters drink imported beer. People in others drink only domestic. Folks in some clusters read the New York Times. Folks in others read the New York Post. People in still others do not take a newspaper at all. If your product can be shown to be particularly attractive to people in certain clusters, you can use this in your retention program. In a previous book, I described how the Globe and Mail in Toronto made highly profitable use of cluster coding to determine who read its newspaper and who didn’t. With that knowledge, it was able to concentrate its telemarketing on the correct postal walks, saving millions of dollars per year in acquisition costs.

Should you use cluster coding? First of all, remember that cluster coding works only with consumer products. Second, consider whether your product appeals particularly to a certain class of people. If you think that it does, spend the money to have cluster coding affixed to 20,000 of your customers, ranked by revenue, and see if the clusters show you anything worthwhile. Any good service bureau can have cluster codes affixed to your file for about $120 per thousand, so your 20,000 will cost you a minimum of about $4000. In most cases, the clusters won’t show you anything at all. If they do, you can use them to make some money. Let’s assume that you have been sending direct mail to about 1.2 million customers per month, for which you spend about $650,000. Table 13-2 shows what your results might look like.

Table 13-2 : Results before Using Cluster Selection

Nonselected direct mail

Cost

Sales

Rate

Cost per sale

Clusters with over 20% penetration

219,499

$111,944

7,463

3.4%

$15.00

Clusters with 5% to 19% penetration

452,991

$231,025

9,513

2.1%

$24.29

Clusters with less than 5% penetration

601,223

$306,624

4,810

0.8%

$63.75

Total customers

1,273,713

$649,594

21,786

$29.82

It is costing you about $30 to acquire a customer. This is your control. Now let’s try to beat that. We are going to select prospects for mailing mainly from the correct clusters, with some others added to make sure you are right. Your mailing next month could look like Table 13-3.

Table 13-3 : Results after Using Cluster Selection

Selected direct mail

Cost

Sales

Rate

Cost per sale

Clusters with over 20% penetration

950,000

$484,500

32,300

3.4%

$15.00

Clusters with 5 to 19% penetration

200,000

$102,000

4,200

2.1%

$24.29

Clusters with less than 5% penetration

50,000

$25,500

400

0.8%

$63.75

Total customers

1,200,000

$612,000

36,900

$16.59

Cluster coding

1,200,000

$144,000

Total costs

1,200,000

$756,000

36,900

$20.49

You are still mailing 1.2 million pieces, and you are spending about $100,000 more than before, as your costs include $144,000 for selecting prospects by cluster. Even so, your cost per sale is down by $9, and you have gained 15,000 more customers on the same 1.2 million mailing. This shows the value of cluster coding, if it works for you.

Why did you do any mailing to clusters where your penetration was less than 20 percent? That is a good question. Self-preservation is the answer. As database marketers, we always have to justify what we do to management, which questions everything. “Why did you spend this $144,000?” your managers will ask. “Because I knew from previous tests that cluster coding would work,” you answer. “Then why did you mail to prospects in the lower-performing clusters?” they will ask. You reply, “Because you would never have believed me if I hadn’t tested some of these lower-performing clusters. Now you do.”




The Customer Loyalty Solution. What Works (and What Doesn't in Customer Loyalty Programs)
The Customer Loyalty Solution : What Works (and What Doesnt) in Customer Loyalty Programs
ISBN: 0071363661
EAN: 2147483647
Year: 2002
Pages: 226

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net