The Sweet Spot

Some people can never get enough chocolate. But that doesn't mean they'll order it from a catalog — especially one that arrives too often at certain times and not enough at others.

That was the dilemma Harbor Sweets faced two years ago when it turned to analytics.

The chocolatier has always generated most of its business by mail. It focused on the September-December fall/holiday period, sending four catalogs to the 130,000 names on its file, according to vice president of marketing Billie Phillips.

The problem was that customers were inadequately served for two-thirds of the year. That led to attrition.

So Harbor Sweets tried Longbow, an on-demand analytics program from Loyalty Builders LLC, to identify those likely to fall away. The company entered several criteria, including recency of purchase and amount spent, and ranked customers accordingly.

It then selected 400 likely prospects, a number chosen based on the test's budget. Each received a tailored letter with a special promotion: Free shipping for any order placed during the next two and a half months, a period that included Mother's Day and Father's Day.

The mailing notched an immediate 9% response. “And these were people who hadn't bought for quite a while,” Phillips says.

Then another test was run, suppressing all 400 names from most of its fall mailings. “If they received a catalog at all, they probably would have gotten only one,” Phillips says.

It worked. By the end of November, 27% of these customers had made a purchase.

But Harbor Sweets also learned what it was doing wrong.

“We identified an overabundance of marketing during the holiday season, and a lack of marketing in the shoulder seasons,” says Ian Watters, Loyalty Builders' vice president for client services.

Fall mail volume has since been cut. Mailings were limited to four groups of customers — some 1,000 during a 2006 test, and 8,000 the following year.

Customers in the test panels were chosen based on how the catalogs influenced their buying habits. (This was inferred from the timing of previous orders and catalog mail dates.) In both cases, a test group received all four books, while the other three groups got one less than their usual quantity.

And the result? “We saw the same purchase patterns we'd seen in years past despite the reduction in mailings, which indicates we were mailing too many catalogs to that group,” Phillips says.

That doesn't mean the company's volume is going down. The goal of the analytics project was not to reduce costs, but to reallocate mailings for greater effectiveness.

“We definitely aren't going to stop something that works,” Phillips says. “The printed catalog is a core means for us to communicate. What we'll be doing is saying, ‘How can we make sure the catalogs we do print and mail are reaching the people who'll be most likely to [use them]?’ More isn't necessarily better.”

For more on ROI, subscribe to the MarketingROI newsletter by Richard H. Levey at http://subscribe.chiefmarketer.com/subscribe.cfm.