The General Data Protection Regulation (GDPR) came into effect in the European Union on May 25. The regulation is meant to protect consumer privacy and consumer data, as well as the processes that use that data to make decisions about consumers.
Marketers may think that if their data practices are compliant, then they can move forward without worrying too much about GDPR or e-privacy, but that's not the case. New technologies and techniques that use consumer data must also be put through the privacy lens. That includes the complex, such as AI and machine-learning, but also processes as commonplace as campaign targeting.
The Right to Opt Out of Machine Processes
The GDPR regulation requires consent from consumers to use their data, but it also requires consent for "profiling," which the regulation specifies as a "procedure which may involve a series of statistical deductions...often used to make predictions about people."
That includes any technical process that uses machine-learning, algorithms, or even simple rules to put people into buckets or make decisions about them. Examples cited most frequently include processes that automatically qualify people for certain products, such as credit cards.
Although profiling people in a way that hurts their ability to obtain credit, or even a medical procedure, has the most unsettling outcomes, profiling has far-reaching consequences that should be considered by marketers selling much less sensitive products as well.
A typical retail marketer will use consumer shopping data to determine which product image to include in an email marketing campaign. For instance, people who shopped for patio furniture might be shown an offer for an umbrella, whereas people who shopped for a new doormat might be shown an ad for a new mailbox. If you are using that type of "profiling," you probably need to obtain permission.
The Difference Between Targeting and Discrimination
Targeting is an important element of digital marketing. It increases relevance to consumers while decreasing wasted campaign messaging and media spending. However, not all targeting is equal in the eyes of the EU privacy regulation. Even in companies that do obtain permission to profile, marketers must continue to be aware of every campaign and sensitive to "the safeguards aimed at ensuring fairness, non-discrimination, and accuracy in the profiling."
Using profiling to determine whether someone might want an umbrella or a mailbox is one thing, but consumers may see such forms of targeting as discriminatory if targeting is based on unintended factors, such as income or race, which are much more sensitive.
For example, imagine that a machine-learning algorithm determines a certain target demographic prefers a fast-food-value hamburger meal whereas another demographic prefers salads. If the resulting recommendations target a lower-income group or race with the hamburger, that could be seen as discriminatory because the hamburger is a less-healthy item.
More Accuracy Is Good For AI...and for People, Too
The GDPR push for more accurate profiling provides marketers with the chance to evaluate the source of their data as well as the processes they use to target and profile consumers. Deloitte recently found, for example, that only 29% of third-party data vendors were accurate more than half of the time.
Lack of accuracy in data can be remedied by testing partner vendors frequently and by working hard to collect and refresh first-party data more often. That increases relevance and decreases third-party data costs—a win-win.
Deploying more accurate and precise AI algorithms should be a focus for marketers. The more you can pinpoint valuable data points, the less you need to rely on having massive volumes of data. Ultimately, GDPR and regulations like it are great for AI, marketers, and consumers alike—because, in a quest for relevance, many relationship marketing programs are now likely to be as off-track as the well-known Gillette targeting debacle.
Over time, the new GDPR rules should align the best interest of AI, marketers, and consumers. In the near term, though, marketers will find that some of their marketing goals need to be moved away from short-term goals of scale or sales toward longer-term goals, such as lifetime customer value. Those short-term goals are what inflate the value of third-party data, make AI less accurate, and annoy—or even discriminate against—consumers...
So, we can all say, "Good riddance."
You may like these other MarketingProfs articles related to Segmentation:
- Switching From Product-Centricity to Customer-Centricity With Personas
- How to Conduct Effective Audience Analysis in Six Steps
- Six Key Criteria for Market Segmentation [Infographic]
- Smoking Brisket and the Customer Experience: Art and Science With Christian Selchau-Hansen on Marketing Smarts [Podcast]
- Target-Audience Segmentation: Why You Need It and How to Do It in Five Easy Steps
- Five Segmentation Gaffes (And How to Avoid Them)