The advantages frequently cited for the use of AI in marketing are based on the assumption that all people online are potential customers. Companies placing ads are interested in transforming potential consumers into real customers – a process known in marketing as “conversion” – by making them aware of their products and convincing them to make a purchase. The companies placing the ads are interested in motivating their potential customers to become real customers – by making those potential customers aware of their products and convincing them to make a purchase. From the perspective of these companies, whether users consent to the use of their personal data or agree to receive ads is of secondary concern. They can, of course, use adblockers and adjust their cookie settings to protect themselves. But they are often unable to decide for themselves the degree of access to their personalized data they are willing to allow.

Our survey showed that only between 13 and 22 percent of users find personalized advertising to be “interesting,” “helpful,” “welcome” and “trustworthy.” By contrast, around 45 percent find it to be “uninteresting,” “annoying,” “unwanted” and “manipulative” (Figure 1). In short, only a small minority of users have a positive view of personalized advertising.

Around half of those surveyed take measures to protect themselves from the collection of personalized data (Figure 2). Some 50 percent of respondents use ad blockers and avoid (at least to a certain extent) digital services that intensively collect data. Less than half of respondents use specialized data-saving search engines and browsers. The majority of those survey participants who do not use such data-saving options said they weren’t aware they existed, don’t know how to use them, find them too complicated or doubt they are beneficial. In sum, around half of all survey participants report taking no measures to protect themselves against tracking and data collection even though they would like to have such protection. Which means that, at a minimum, methods of ensuring such protection must be made more accessible, more effective and more visible – or companies could forego the collection of personalized data.

On the flip side of the coin, the company representatives interviewed for this study say they have limited room for maneuver if they want to take advantage of the personalized advertisements made available by the big players in the industry. “If you want to participate in the market, there are certain things that you simply cannot avoid. […] For example, one of our most important [revenue] sources is Google as an advertising medium. And there, we have no influence. We simply have to go along with the market, otherwise we wouldn’t be present to the degree we currently are if we didn’t participate,” says the head of marketing for an online mail-order retailer.

The companies interviewed for this study are primarily focused on issues pertaining to data protection because legal mandates, such as the EU’s General Data Protection Regulation (GDPR) require them to do so, and they would like to avoid violations. It’s not just 80 percent of the households surveyed that believe the use of AI for the personalization of online advertising produces at least some disadvantages for consumers when it comes to data protection and privacy – a similar share of companies do as well (Figure 3).

Those companies that expressed a greater focus on sustainability during their interviews for the study noted their intention to approach customer data with a greater degree of responsibility. A representative of one such company said during the interview that they only collect data that is directly relevant for marketing purposes and only serve ads sparingly and in a targeted manner. Such strategies, though, are not widespread. “I would tend to see the advantages [of personalization] or the opportunities it opens up. For users, I don’t see much of a danger for the time being,” said the marketing director of a smaller online marketplace.

The technology sector’s energy consumption is attracting critical attention from many directions, but such concerns generally don’t play a significant role in the day-to-day of online marketing. That lack of attention to sustainability is further encouraged by the fact that using technologies based on AI and on the findings of data science are constantly becoming cheaper and easier to use. That in turn has led to the industry’s rising energy consumption. “As the technology for computing becomes cheaper, there is a tendency for being less mindful of the performance of certain algorithms. Cloud providers make it very easy for you to scale up the resources if you need to. So then, if you want, you can have a very energy-intensive algorithm and just run it, and it will run fast,” a data scientist from a marketing agency said. Effectiveness, speed and costs are more important than environmental efficiency, he continued. “It’s no secret how energy-intensive machine learning is. But I hardly think that there are attempts to improve energy efficiency to benefit the environment, not even from those who are environmentally aware.”

Both small and large service providers do offer to calculate carbon footprints. But our interview partners doubt the effectiveness of such measures. “Google has some services that allow you to calculate the CO2 spend that you have while running the services. […] There are some data sources where they put a little green leaf that says renewable energy has been used to produce them. But again, it’s not very transparent and it’s very easy to greenwash.” Furthermore, it is difficult to force market leaders like Google, Microsoft and Amazon to improve transparency and become more sustainable. That limits the options of companies that rely on advertising but would like more environmentally friendly options.

Further environmentally damaging effects can be produced when advertising results in the sale of more products and, by extension, the consumption of more energy and resources. It is difficult to quantify the degree to which this effect impacts the environment. But companies naturally hope that advertising will boost their sales and they don’t critically examine that goal. The immense amounts of money spent on personalized advertising suggest that such advertising strategies have the potential to significantly increase consumption. One marketing-team member at a mid-sized mail-order company notes: “[Through marketing], needs are met that 20 years ago we didn’t even know we might have one day. For me, that is evidence that marketing produces demand.” The use of AI for the personalization of advertising serves to effectively convert the interests and preferences of users into purchases by activating new needs.

Further environmentally damaging effects can be produced when advertising results in the sale of more products and, by extension, the consumption of more energy and resources. It is difficult to quantify the degree to which this effect impacts the environment. But companies naturally hope that advertising will boost their sales and they don’t critically examine that goal. The immense amounts of money spent on personalized advertising suggest that such advertising strategies have the potential to significantly increase consumption. One marketing-team member at a mid-sized mail-order company notes: “[Through marketing], needs are met that 20 years ago we didn’t even know we might have one day. For me, that is evidence that marketing produces demand.” The use of AI for the personalization of advertising serves to effectively convert the interests and preferences of users into purchases by activating new needs.

When asked if they were aware that AI is being used to personalize advertisements on the internet, only 55 percent of survey participants from private households responded with “yes.” Only 29 percent feel that they are well informed about the kinds of personal data collected, when it is collected and what companies are doing the collecting. Over half of company representatives and survey respondents are aware of the dangers posed to data and privacy protection. But many companies and private citizens are unsure about the risks posed by AI to the environment or to the formation of political opinions and to democracy (Figure 3).

This lack of knowledge, combined with the lack of transparency in marketing practices, explains why actors in both civil society and in science are demanding that data protection not be reduced to individual decisions, but that it be defended as a legally protected fundamental right. Organizations like Germany’s Digitalcourage, the Chaos Computer Club (CCC) and Netzpolitik, in addition to companies like Mozilla and political actors like the European Commission have for years been pointing to the opacity, surveillance, manipulation and discrimination of the advertising industry. “The biggest risk is the possibility for manipulation. […] The more detailed the information collected about people […], the easier it is to manipulate them or evoke a certain behavior,” says a representative of an association that promotes digital data protection.

Many experts from the worlds of civil society, academia and politics are thus demanding the labeling of Artificial Intelligence in advertising campaigns or a general ban on the use of personalized data in online advertising. Personalized online marketing must be regulated and restrictions on advertising discussed to protect both our democratic societies and the environment.

Frick, V., Marken, G., Schmelzle, F. & Meyer, A. (2023). “The (Un-)Sustainability of Artificial Intelligence in Online Marketing. A Case Study on the Environmental, Social and Economic Impacts of Personalized Advertising .” IÖW Schriftenreihe 228/2023. ISBN 978-3-940920-33-1.