Research Examines True Cost Of A Solicitation

September 15, 2007       Marla Nobles      

“Solicitations are investments. Sending a solicitation is an investment with a quantifiable return,” said Edward Malthouse, Ph.D., an associate professor at Chicago’s Northwestern University. But according to Malthouse, quantifying the worth of a direct response solicitation is not always easy.

Malthouse began thinking about this rather commonplace, yet complex, issue a few years ago. What followed was a stream of research, the findings of which were unveiled at this year’s Direct Marketing Association Nonprofit Federation (DMANF) conference, in New York City.

Malthouse, who teaches integrated marketing communications at The Medill School, Northwestern University, conducted the study using data provided by three national nonprofit organizations that are DMANF members. The study was financed by the Don Kuhn Fund, which since early 2006 has funded research and scholarships in honor of the late direct-marketing guru.

“The purpose of the study is we want to estimate the true value of the marketing solicitation,” Malthouse said during a presentation of the study’s findings at the DMANF conference, held at the Waldorf=Astoria Hotel. He said the contribution of the study was to show how to estimate the return on investment (ROI) of a donor to improve targeting.

According to Malthouse, for years nonprofits have looked at the “simple math.” For instance, if a nonprofit mailed 100,000 donors and generated $100,000 from that mailing, each donor was worth $1. “But that’s what I call short-term revenue,” said Malthouse, “and that undervalues a solicitation.” Malthouse said organizations should account for the donation along with what follows: the long-term and auxiliary effects.

Long-term effects The long-term component, said Malthouse, is the expected incremental customer long-term value (CLV) due to contact, where the recency, frequency, and monetary value (RFM) status changes if you get a response.

According to Malthouse, a scoring model – a data-mining model used to predict behavior based on other information available on a prospect – is used to estimate the short-term component. It ranks people from best to worst on their responsiveness to an offer as well as to estimate the probability of a response and donation amount. To estimate the long-term, an organization must look at CLV*, which estimates CLV as a function of RFM and c, the vector indicating future contacts.

“Modeling CLV* is difficult because most organizations don’t know ‘c,’ which will depend on responses to future mailings,” said Malthouse. “The problem is, they don’t have a crystal ball.”

Another way of estimating long-term value is the scoring model approach. “Turn back the clock several years. For example, pretend today is December 1, 2003,” said Malthouse. “Then we can compute the exact long-term value over a multi-year period by summing the discounted actual donation amounts from December 2003 through the present.”

Next, predict CLV from what was known about the person as of 12/2003, for example, RFM as of 12/2003. Malthouse said all of this can be done using commercial statistical software packages and/or employing the expertise of modelers who have experience with multiple regression and building scoring models. There are agencies and consulting firms that provide expertise as well.

Malthouse argued nonprofits should place more value on the long-term component, particularly the value of moving a donor to a more loyal donor state. “By that I mean, you’ve donated more recently now,” said Malthouse. “We all know that the more recently you’ve donated the better you’re going to perform in the future.” Likewise, the more times a person donates the more likely that person will donate again.

Malthouse said he estimated the value of moving someone to a more loyal donor state at between 3.5 to 8 cents per donor for the three participating organizations. “That’s big money,” added Malthouse, “because you’re multiplying that across large numbers of customers.”

Moreover, an additional 3.5 to 8 cents can turn a loss into a gain. “Such that if we expect to make 47 cents to mail you, and the mailing costs 50 cents, if you add that 3.5 to 8 cents on top of that, it’s now profitable for me to mail you.”

Managers who decide mailing depths based solely on short-term profits will under-mail their files, added Malthouse.

Auxiliary effects “The other new feature that people need to account for is what I call ‘auxiliary effects,'” said Malthouse, which is the expected incremental donations to other channels or mailings within the channel. Advertising often has cumulative effects and repeated exposures have positive effects to a point. He cautioned that donating to one mailing might cause someone to not donate to the next mailing.

Additionally, donors might prefer to donate through other channels. For example, a person might receive a direct mail solicitation and instead go online to make a donation. This bodes the question: Would that donor have gone to the Web site without the mailing? “And if the answer is no, then that direct-mail piece ought to get some credit for generating the online donation,” said Malthouse.

A person could receive a mailing and choose not to donate at that time, but due to the recipient’s heightened awareness of the organization and cause, he is more likely to respond to a future mailing. “The auxiliary component is a very tricky one to estimate for statistical reasons,” said Malthouse. “If you look at the way a lot of nonprofits in general manage their mailings — customers of the same type often get the same mailings — that makes it really hard to make a valid comparison.” For this reason, he said, “we have to test into it.  That was my message.”

One finding, said Malthouse, was that there were some synergistic effects before the holidays. “So you’d want to test ramping up some of those pre-holiday mailings, sending them to more people,” he said. “While relying on a single mailing in the past to make your decisions about the future, you look at what happened to multiple mailings in the past and you average your findings from those to predict what’s going to happen in the future.”

Mailings in March, April and May cannibalized the June mailing. The odds of those who received the April mailing responding to the June mailing, for example, were 66 percent less than those who did not receive the April mailing. “The implication here is, it looks like you can increase mailing volume in a build-up to the holidays, but then you want to cut back because there’s cannibalization after that.”

Malthouse said his method accounts for year-to-year changes, including natural disasters, different economies, etc.  NPT