Marketers have a love-hate relationship with data. On the one hand, they can’t get enough of it, and they’re always asking customers for more. On the other hand, it can be difficult to manage and, sometimes, cause marketers to get fined or worse if acquired or used unethically.
But, beyond the obvious, what constitutes unethical use of customer data? And perhaps more important, why should marketers care about how their data scientist colleagues are using customer data? Marketers are neither statisticians nor philosophers. And some marketers’ own data collection and usage practices are less than honorable.
“Advertising? Ethics? Fucking weird,” said Tim Rich, director of data science for interactive agency Publicis North America, during his session at SXSW in Austin.
During his talk, Rich explored the topic of data ethics and argued why and how data scientists should adopt their own code. Here’s a breakdown of the reasons marketers should care about the topic, where data scientists should start in creating their own code of ethics, and how the industry can enforce it.
1. Why should marketers care about data and ethics?
According to Rich, marketers should care about the ethics around data use for two reasons: First, they spend a lot of money on marketing; second, they have a lot to lose.
Marketers invest heavily on advertising and other marketing initiatives with the aim of attracting customers. In fact, data released by eMarketer in September 2015 shows that total worldwide media ad spend is expected to reach $606.9 billion by the end of this year and $719.2 billion by 2019. Yet, more and more customers are turning to ad blockers to shield themselves from unwanted promotions. According to “The Cost of Ad Blocking” report by PageFair and Adobe, the number of global ad blocker users grew 41% from Q2 2014 to Q2 2015, and the global cost of ad blocking will reach $41.4 billion by this year.
“There’s a lot of bullshit on the Web that we’re responsible for,” Rich said.
For some, data scientists are the glimmer of hope to help organizations succeed. After all, data science powers many aspects of society today, Rich noted, from media to politics. Yet, there are no standards or guidelines on how data scientists can use this information ethically to benefit marketing. Rich asserts that data scientists need a code of ethics to establish credibility, provide a starting point, and galvanize disparate data practitioners.
2. What should the code of ethics look like?
Thankfully, data scientists have centuries-worth of history to consult to help assemble their code. From the Egyptian Ma’at to the American Statistical Association, codes of ethics, as Rich pointed out, have stood the test of time. And all of them, he added, have included two fundamental elements that data scientists should adopt when creating their code: inward goals and outward goals.
Inward goals, he explained, provide governance for practitioners. He said they build a common purpose, establish professional behavior, and deter unethical behavior through sanctions and internal reporting.
Outward goals protect vulnerable people who could be harmed by the profession’s activity, he added. They serve as tools to resolve disputes between members and non-members, create institutions resilient to external pressures, and establish the profession as a moral community worthy of autonomy—something , Rich noted, marketers are fighting for today.
3. How can the data science industry enforce this code?
Once data scientists create a code of ethics, how can they ensure that it’s enforced? “Well, you don’t,” Rich said. This may seem counterintuitive, but proper enforcement would require an accepted consensus, Rich said, as well as specific boundaries and expectations, and self-affirming social control. Challenging, yes, but not impossible to develop over time.
“This is why we all wore pants today,” he said. “It’s a self-affirming social contract.”
He added that the code writers would need to consider the current and future needs of the group and establish a cultural identity that would be recognized by group members and industry-wide.
Of course, an organization creating an industry-wide code of ethics would have to consider the moral hazards surrounding data science, too. The abstract connections between data and the people it represents is one hazard, Rich said, as is the notion that many of marketers’ data-driven digital initiatives can personally affect the people they’re marketing to. He also cited how unintended algorithmic consequences that are almost never identified or explored can be hazardous, as is the blasé idea that an algorithm never hurt anyone.
“Data science is steeped in moral hazard,” Rich said, “and no one likes to talk about this.”
4. How can the industry overcome these hazards?
The key, he said, is to create a code that balances utility while also maintaining control of the data. To do this, he added, marketers and data scientists need to answer some questions surrounding three key areas: community, identity, and privacy.
When it comes to community, he said, those in the industry need to figure out the following: Once data is used, how is it, as well as the sensitive analysis around it, discarded? How can marketers and data scientists protect people affected by their analysis from negative consequences?
Issues related to identity are centered on ownership, validation, and data correction. In terms of ownership, could there be a centralized personal “data safe” or repository (like a bank), Rich posited, where customers could store their information and access it when they need to? He noted that marketers and data scientists need to consider how validation could affect access, privacy, and safety, as well as what the mechanisms they could provide to allow people to correct faulty information?
“Very few people in companies actually have [processes] in place to correct data,” he said.
The issue of privacy is especially complex around the idea of ownership. If a marketer buys third-party data, Rich noted, who owns it? There are also persistent questions around Do Not Track; for instance, he cited the relationship between the expected privacy of Internet browsing and delivering relevant advertising.
Once the industry starts answering these questions, Rich said, it can begin to move forward.
“All of these together, I believe, form the basis of…starting to think about a conversation of data and ethics,” he said.
4. How can the code writers overcome any internal issues?
In attempting to write an industry-wide code of ethics, there will be internal issues that the code writers will need to work out. Rich cited as an example that the code cannot be built based on personal ideals of right and wrong. Instead, he said it needs to be born out of ideas general enough to span companies, cultures, and continents.
He added that the code should not exist within a formal business, and people should not be able to make money off of the code.
Finally, he argued that the code should not be created by a small group, but should present an opportunity for democracy.
5. What are the benefits of creating a code of ethics?
Creating this proposed code of ethics won’t be easy, but Rich argued that it will be worth everyone’s while. One benefit: It’s good for business. Rich said that the industry has been so focused on trying to monetize data and generate value from it, that it stopped using it come up with hypotheses, test ideas, and confirm preconceived notions. Another is that often companies considered part of the “moral high group … sell more shit,” he said—referencing TOMS’s business model and Box Tops for Education.
6. Where should the industry start?
This revolution isn’t going to happen overnight. But Rich said that marketers and data scientists can take the first step simply by having conversations about data and privacy with their colleagues. He’d also said that he’d like to see more data ethics classes taught in schools—ones that take a classical approach and reference Aristotle and Plato to solve modern issues.
Although the road to Rich’s revolution may be long, it’s an uprising that marketers and data scientists need to lead head on. After all, it’s the ethical thing to do.
