Most of marketing analytics is designed to help get the right offer to the right person. That is why it’s sometimes hard to consider that what marketers think is so cool – the data segmentation and modeling that enables those relevant offers to be matched to likely recipients – could be considered creepy by others.
This divide is in some ways caused by the language of marketing analytics, and a recent DMA Analytics Council “Town Hall” conversation kicked off a discussion of what to do about closing the gap. The Town Hall, “A New Language for Predictive Analytics,” was conceived and led by Peter Zajonc, Senior Director, TS and TSP Analytical and Modeling Services, Epsilon, and the Membership Chair of the DMA Analytics Council.
“The work that Analytics professionals do – like predictive analytics – is sometimes misunderstood by the FTC and other regulators, and helping them understand is something each of us can support,” Peter said. Inspired by the testimony of DMA to the FTC on “marketing scores” earlier this year, Peter and other Council members are working toward developing some new language recommendations that will celebrate the good work of data-driven marketing, while calming the fears of regulators.
Rachel Thomas, VP, Government Affairs, DMA, participated in the Town Hall and explained that there are two worlds and two languages between marketers and policymakers. Congressmen and senators are all consumers, she said, and when they hear words like “scoring” and “modeling,” they get concerned.
“Credit ‘scores’ are used to make life-changing decisions, like if someone gets a job or healthcare or a mortgage or a place to live,” added Rachel. “Policymakers assume that if marketers call something a ‘score’ then we also have a big important life-changing impact which could be potentially harmful and thus should be regulated.”
Don Hinman, former SVP, Data Strategy, Epsilon, described data as the “fuel” used to create these marketing scores and models in predictive analytics. Don said, “Consumers are largely unaware of these scores. This can raise a variety of privacy and protection fears. But the goal of the scores is always to determine the right group to market to.” Don even suggested a new “Hippocratic Oath” for marketing analytics folks, “First, do no harm.”
Such concerns underscore the importance of transparency, which can be largely accomplished through purposeful, accurate language. Peter said, “By paying particular attention to language, marketers can better educate consumers and policymakers about marketing and analytics.” The Town Hall participants discussed several examples of sensitive marketing language and how these can be revised to build trust in the industry.
Nicole Tachibana, Privacy Manager, Epsilon, pointed out that it is important to be sensitive to offensive labels for marketing segments or lists. For example, if a marketer’s intention is to provide a group of low-income consumers with affordable or discounted products, then the marketer could unintentionally engender distrust among policymakers or the public by applying a marketing label to that group that could be perceived as negative.
“If outsiders can’t get past the first checkpoint – the name of a segment – and assume we’ve done some sort of discrimination, it’s very distracting to the conversation,” said Nicole. “If we want to move past this first step, we need to take the focus off these negative points and focus it on the use of data. We start off on the right path when we label and present our use cases in a way that is consumer sensitive.”
Peter claimed that action is needed at all levels, especially because analytics professionals will not always have control over the use of a model or a label. “The work leaves our hands and goes somewhere else,” said Peter. “Look at your standard deliverables and see if the language is appropriate.”
Another word that the Town Hall discussed at length was “discrimination,” which, for marketing analytics is merely descriptive of a good end result because it connotes that an audience has been correctly segmented for marketing purposes. Malcolm Houtz, Analytics Director, Alliant, said, “It never occurred to us that it [the word ‘discrimination’] had any negative connotation. We’ve got to all have that level of awareness.”
Evan Balzer, Vice President, Business Development, Guideposts and the Ethics Chair for the DMA Analytics Council, agreed. “It doesn’t matter what we think—it’s how it is interpreted. Not just on the federal level, but also on the state level. Awareness will beget actions,” Evan said.
Don had a slightly different view on the word “discrimination,” and he summarized a conversation he had once with officials from the FTC explaining that the word “discrimination” has multiple meanings and, when used by marketers who develop these models, is a statistical phrase to find the right people to market to and is not intended to harm individuals.
Keith Bergendorff, AVP, Analytical Services, Publishers Clearing House, was optimistic about the future of predictive analytics and the ability for marketers and policymakers to reach an understanding. “If you represent yourself as a practitioner and you explain the purpose of your work clearly, then even if some of a model’s variables are arcane and difficult, you can gain a lot of trust from regulators who may be skeptical but are genuinely trying to figure it all out.”
Rachel of the DMA agreed, “We need to calm and educate policymakers about the fact that we intend only to serve better offers. If we don’t draw the line, then someone will draw it for us.”
The Town Hall resulted in some good advice for all analytics professionals and every marketer to help the entire industry build trust with policymakers and consumers:
How do you feel about the language of predictive analytics? Let us know in the comments section below – and also provide any requests for future Town Hall topics. If you would like to join the DMA Analytics Council, please visit our website or just contact Stephanie Miller.