Today in Digital Marketing - Special Episode: "Genetic Predisposition" — The Next Targeting Option?

Starting point is 00:00:00 Hello, today's episode is a special edition as I'm out of the country for agency business until Tuesday night. In those four days of absence, though, we are bringing you some deep dive interviews with marketing scientists. These interviews are usually exclusive to members of the Premium Podcast or our Premium Newsletter, but we're bringing you four of these during my absence. If you like what you hear and want to learn more about the premium podcast, just tap go premium in the show notes. Our regular marketing newscasts return when I'm back on Wednesday. Enjoy. Do you have business insurance?

Starting point is 00:00:33 If not, how would you pay to recover from a cyber attack, fire damage, theft or a lawsuit? No business or profession is risk free. Without insurance, your assets are at risk from major financial losses, data breaches, and natural disasters. Get customized coverage today starting at $19 per month at zensurance.com. Be protected. Be Zen. In 2018, Spotify had an interesting marketing campaign. It offered its users the ability to upload their genetic data, which Spotify would then turn into custom playlists that matched their genetic ancestry. A few months later, Mexico's national air carrier, Aeromexico,

Starting point is 00:01:14 launched a DNA discounts program, offering discounted flights to the country. The more Mexican you were, genetically speaking, the more of a discount you'd get. Welcome to the new world of genetics marketing, one where we might soon see genetic predisposition listed in Facebook's ads manager as a targeting criteria. It's a topic Rami Davia studied. He's an assistant professor of marketing at the Wisconsin School of Business. He and his colleagues recently published a paper called Genetic Data, Potential Uses and Misuses in Marketing. And he joins me now. Professor, welcome. Thank you.

Starting point is 00:01:51 So first, how does this work logistically? Someone buys a consumer testing kit, then they upload the results to like a social media platform. So they buy the testing kit, which is usually just a plastic tube that you fill with spit, and then you seal the tube and ship it. Once the tube is shipped, it's analyzed by a lab, and the data is released to your provider, so 23andMe or Ancestry.com,

Starting point is 00:02:18 and then you have this data available to both you and your provider. What's been your understanding with consumers' comfort level with that? So far, I think the consumers are mostly uneducated. They might see it just as some data about themselves, like their shopping behavior online or their purchase, whether it's online or offline. They don't necessarily understand that once a company has their genetic data, they have something that's informative about virtually everything in their lives, like the educational attainment, what they like to eat, at what time they wake up,

Starting point is 00:02:55 like all these things can be partially inferred from DNA. And so when you send your DNA to say, am I tasting soap when I eat cilantro, you don't necessarily know that this DNA can be used to answer a lot of other questions that could be used for marketing purposes. Like what? Taste is one of these. Like I mentioned, cilantro, you have other, like 23andMe will tell you if you prefer vanilla or chocolate. We ran actually a study on general tastes of food, and we can see that virtually all tastes, so spicy, salty, savory, they can be – the preferences regarding these tastes can be partially predicted by genetic code. But you also have everything related to your circadian rhythm. So as I mentioned, if you wake up early, if you like to do sports, if you study a lot, if you're more of an outdoor person, if you like gathering, or if you prefer to stay by yourself, I can go on forever because almost any behavior we have data on, we could predict it to some extent with genetic data

Starting point is 00:04:08 and usually the R-square. So the percentage of variation that we can explain based on the DNA for behaviors is between 15% and 25% for all of them. But that requires very large samples. So it's not easy to run these studies. And a big company that has 30 million customers can do that. For us scientists, it's a bit harder. You know, the timing of your study, I thought, was really interesting because we marketers are in a kind of time and space right now where partially because of privacy, partially because of legislation, we are getting fewer and fewer targeting abilities. Facebook just this week removed a whole bunch of targeting abilities.

Starting point is 00:04:56 So are these genetic predictive estimates good enough to predict that someone might be predisposed to purchase a specific product? Like could a coffee shop, for instance, use genetic data to target consumers who have a genetic potential to like espresso? So coffee shop is a specific case because there is a very specific genetic variation that change your sensitivity to bitterness. So for coffee, we might have better actually predictive abilities

Starting point is 00:05:23 and other traits that rely on thousands of variations. But for most behavior, DNA by itself is not predictive enough to do individualized targeting. It could be good for populations, but not for individuals. But when we ran other studies where we added demographic data or behavioral data, we found that a prediction that could be already good based on demographics and past behavior will be even better at adding genetics. So once you combine all these data sets, that's when you have a really good targeting accuracy. So sort of layering it as an additional layer on top of it to tweak it? Yes. And since genetic information is somewhat, as I understand it, at least shared between family

Starting point is 00:06:13 members, could marketers use this to find out who else in the household might want to buy a specific thing? Yes. So technically, I did a test and they have 50% of my mom's DNA with my data. If they found out that I was lactose intolerant, they could actually try to target my mom with soy milk or almond milk or alternative dairy product. The problem here that we're facing is not really technical, it's more ethical. My mother never consented to give her DNA to have it mined. So actually, she is in Europe. So she's protected by GDPR on that regard. But in the US, there is no protection regarding family members. And actually, this data has been mined not necessarily for marketing, but to find the Golden State killer. It was found through

Starting point is 00:07:06 relatives who shared their DNA on ancestry platforms. And they could say, oh, this killer we're looking for is a member of that family, that family, and that family. And by triangulating, they found who it was. So this is completely possible. How could this affect market research? You note in your paper that our current methods have got flaws, basically, when it comes to some topics, like flaws in terms of the scientific method, right? Studying the relationship between someone's consumption habits and their long-term happiness, because the proper scientific method would require us to put one set of subjects into a group where we might, I don't know, reduce the nutritional value which, you know, obviously would create a threat to their health. So can genetics play a role in filling some of that market research gap? So for causality, I think the biggest potential of genetic data for marketing is based on correlation. So just predicting you are a member of this group, you are more likely to have these habits.

Starting point is 00:08:04 We're not sure if it's caused by the genes or if the genes are just informative about it. However, there is this very specific characteristic of genetic data that you're born with it. And whatever you do in your life, it won't change until we have genetic kits to modify your DNA available, but that's not authorized by the FDA right now. So since DNA is immutable and is kind of random because if you know your parents' DNA, you got randomly 50% of your mom's DNA, randomly 50% of your biological dad's DNA. So once we know that, we have a natural experiment, basically, that gives you both the randomness and the independence on the side of the DNA. And you can use this as an instrumental

Starting point is 00:08:54 variable. So or if you are in quantitative genetics, you will call this Mendelian randomization to find the causal effect of DNA on some behavior. For researchers looking at causality, it's possible to use DNA, and DNA is actually a very good candidate due to this characteristic of not being modified by external factors like other variables. So the causal direction is very clear. Right now, when marketers go to place an advertisement on, say, Facebook, we get presented a list of targeting options. One of them is interests, you know. So for instance, are they interested in coffee? Are they interested in sports? Do you think that five years, 20 years, one year from now, we will see a marker for genetics where that can be a

Starting point is 00:09:39 targeting option as well? So if you're looking at 23andMe specifically, they mentioned already that their goal is not necessarily to make money by selling kits. Their goal is to become the Google of personalized healthcare. So, in the specific market of healthcare, they do intend to monetize the data. And that's likely by offering you suggestions about what product you could use to improve your health, to prevent some future potential disease, and this kind of thing. So they could actually target you with this kind of data, either through their own platform, or by selling or sharing a mailing list kind of with Facebook and say, we know you have these people on Facebook and we would like to send them this ad about hair loss product because they are more likely to lose their hair as they reach 35, 40.

Starting point is 00:10:36 What surprised you the most about your research, Professor? Oh, the first surprise is how few ethical guidelines there are for businesses. Another surprise was how informative DNA was, which would need. This is why we require more ethical guidelines and maybe regulations. So you have something you're born with that can predict your life 60 years ahead since you're born. Of course, it's a very noisy signal, but it can still have some predictive ability. So when I'm born, you might have some companies saying, oh, when this person turns 18, they are more likely to like coffee. Give me another variable that can predict this so far ahead. So imagine companies starting to target specific consumer

Starting point is 00:11:28 when they are 13 or 14, because they know that when they turn 18, they might develop some attraction to a specific kind of product. So yeah, this is fascinating because it's kind of a unique data and I didn't expect it to have actually so much potential for marketing. Do you think this discovery or this adaptation of this data is generally good for society?

Starting point is 00:11:55 Generally bad for society? It has a lot of good potential. given the current climate of absence of regulation and some firms who show no concern toward ethics or moral, I'm a bit worried, actually. I'm less worried in Europe where it's strongly regulated, but in the rest of the world, it's very uncertain where we are going to, I think. Well, it's fascinating and scary all at the same time. It is. You know, it's interesting, I think, being a marketer because you're always wearing two hats. You know, you're wearing the marketer hat where you're excited by opportunities like this. But we are all also consumers in the marketplace.

Starting point is 00:12:43 And so the other hat that we're wearing is, wow, do these filthy marketers have access to this? This is terrible. So it's an interesting dichotomy. But, Professor, thank you for your time and sharing this research with us. It was a great pleasure. And thank you for inviting me. Remy Davia, he's an assistant professor of marketing at the Wisconsin School of Business. His paper is called Genetic Data, Potential Uses and Misuses in Marketing. You can follow him on Twitter. Thank you.

Today in Digital Marketing - Special Episode: "Genetic Predisposition" — The Next Targeting Option?

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.