Is Targeted Advertising Ethical?
Targeted advertising is a huge industry with a massive branding problem. On one hand, nearly all of the “free” digital products and services used by millions rely on targeted advertising as their primary business model. On the other hand, users are increasingly anxious about how they are being tracked and how that data is shared. How do we resolve this contradiction? Is there such a thing as ethical targeting?
Sometimes users truly appreciate being targeted. Users generally appreciate recommendations that are optimized for their tastes. Everyone can benefit from better targeting in a two-sided marketplace, like a talent marketplace that brings together employers and job seekers, or a dating site that brings together compatible romantic partners. Removing targeting from such applications and platforms would defeat their purpose and undermine the value these applications create for their users.
However, while these examples illustrate the value of targeting, they do not apply targeting to advertising, where relationships among the parties can become more adversarial and ethically fraught.
Digital advertising mostly shows up in two kinds of applications. Search applications often include promoted search results optimized for something other than the searcher’s utility. Ad-supported media (including social media) applications embed ads on the page or the feed. Whether and how these ads are targeted depends on a variety of factors.
Promoted Search Results
In a search application, searchers expect results to be relevant and ranked by their desirability. That desirability can be personalized based on the demographic information and interests in an explicit user profile or implicit signals derived from the user’s past behavior. Searchers benefit when search applications use these signals to improve result quality.
What about promoted search results? By definition, these are not optimized for the searcher’s utility. Still, targeting them can make them more useful to the searcher. To promote shirts, a search application can use the searcher’s gender and size to select shirts the searcher can wear. Moreover, it can choose styles similar to those of the shirts the searcher previously purchased. The promoted results may not be optimized for the searcher’s utility, but targeting does improve them. After all, the alternative of untargeted promoted search results is just as invasive but less useful.
Ad-Supported Media Applications
The situation is different for media applications. Users mostly access media applications to consume content, rather than to search for stuff. In some cases, the content implies the user’s intent, e.g., watching a video about coffee makers suggests an interest in buying one, and advertisers can try to harvest that intent. More often, however, the content does not neatly map to an intent for which there are relevant ads. Instead, advertising is the price users pay to consume content without paying cash.
Again, targeting can make ads more useful, whether based on the user’s explicit profile (e.g., gender, age) or signals derived from the user’s behavior. Many media applications rely on retargeting based on behavior on other sites, e.g., shopping behavior on e-commerce sites. Users find targeted ads more useful (or at least less useless) and do not need to consume as many of them, since targeted ads yield more revenue per ad.
Transparency, Tracking, and Sharing
Promoted search results offer a revenue model to search applications that at least partially aligns with searcher utility. Ad-supported media applications lower the number of ads users consume by improving ad effectiveness through targeting. So where is the ethical dilemma?
One concern is transparency. When a search application presents promoted results, they must be clearly labeled as such so as not to be confused with organic results. Similarly, ads in a media application must be clearly labeled to distinguish them from organic content.
However, the biggest ethical concern about targeted advertising is data tracking and sharing. Users are wary of applications recording and analyzing their behavior, and then sharing that information with third parties. That is why the backlash against targeted advertising focuses on privacy rather than the ads themselves. Users’ biggest concern about tracking is that their private data will be shared with malicious actors.
No Free Lunch
All else equal, users prefer free to not free. But all else is not equal. Most free applications rely on ads to generate revenue, and there is an effective social contract that users pay for the free application with their attention, which the application then sells to advertisers for cash.
Targeting makes that attention more valuable to advertisers, especially for media applications. Moreover, effective targeting for media applications generally relies on retargeting.
So long as most users are unwilling to pay cash for digital products and services, we are stuck with ad-supported models. The question is then whether the ads are targeted or untargeted. Untargeted advertising may improve privacy, but at the cost of devaluing user attention, leading to lower ad revenue and more invasive ads. This is a lose-lose.
As I see it, users have — or should have — two choices: pay cash or accept targeted advertising. If there are going to be ads, they should be transparent. More importantly, users should know what data is tracked and with whom it is shared. And there should be legal restrictions on how this data is used, regardless of whether users consent to its collection.
But in the end, everyone needs to recognize that there is no free lunch.
