Navigating AI: The pitfalls of ad personalisation that marketers must avoid

AI-powered personalisation in the ad-tech industry has indeed brought about a significant shift in the balance between brand-customer relationships and consumer privacy. On one hand, it enables brands to deliver highly targeted and relevant content to consumers, thereby enhancing their overall experience and engagement. However, on the other hand, it raises concerns about consumer privacy and the potential for intrusive advertising practices.

One challenge of utilising generative AI for personalisation is the potential for overstepping privacy boundaries. After all, it is a very thin line. Generative AI techniques, such as natural language processing and computer vision, have the capability to analyse vast amounts of user data to create highly tailored content. While this can lead to more personalised experiences, it also raises concerns about the collection and use of sensitive personal information without explicit consent.

Furthermore, there are challenges related to the transparency and explainability of AI-powered personalisation algorithms. Consumers may feel uneasy about receiving personalised recommendations or advertisements without understanding how their data is being used or why certain content is being presented to them.

In this two-part series, Adgully seeks to navigate through the realm of AI-led personalisation ads, the risks and pitfalls, and more.

The challenge with personalised experiences lies in meeting consumers' expectations for tailored and relevant content, while respecting their desire to share minimal personal data, says Adam Smart, Director of Product - Gaming, AppsFlyer.

“With regulations, laws, and platform adjustments aimed at safeguarding users, consumer awareness around the value and sensitivity of their data is growing, prompting a reevaluation of the information exchange. In 2023, there was a surge in the integration of generative AI across digital platforms and services, promising enhanced customer experiences, cost reductions, and operational efficiency. From text generators such as ChatGPT and Google’s Bard, to image generators like Midjourney and Dall-E, and even the increasing popularity of AI-generated music, each application brings unique use cases along with privacy considerations,” says Smart.

“While AI-powered personalisation offers numerous benefits, the ad-tech industry must navigate the challenges of privacy, intrusiveness, and algorithmic bias to build and maintain trust with consumers. Striking the right balance requires careful consideration of ethical practices, transparency, and a commitment to user-centric design,” he adds.

AI-powered customisation has undoubtedly transformed the landscape of brand-customer relationships, providing unprecedented opportunities for targeted and relevant content delivery, observes Hiren Shah, Founder & Chairman, Vertoz. However, he adds, this enhancement requires a delicate balance in terms of customer privacy.

The challenge, according to Shah, is to prevent the intrusive nature of personalisation, where AI algorithms, particularly generative AI, may unwittingly violate privacy boundaries. “Striking the right balance requires a meticulous approach, respecting user consent, implementing robust data protection measures, and ensuring transparency in data usage. To create and sustain consumer trust, brands must address ethical considerations,” he adds.

AI-powered personalisation has revolutionized brand-customer interactions by delivering heightened user experiences and fostering increased engagement through tailor-made content, says Dhaval Gupta, MD, CMRSL, the parent company of CMGalaxy.

“Despite the positive impact of personalized advertising on perceptions, valid concerns arise regarding privacy, data security, and perceived intrusiveness. As ever, it is important for brands to secure their own data, and more importantly the customers’ data. Therefore, brands should invest in relevant technology, or find strong tech partners that guarantee data security and privacy protection,” Gupta says.

Gupta observes that overly personalised ads could inadvertently create a feeling of surveillance, potentially affecting user experience and straining brand-customer relationships.

Consequently, he adds, there’s a recognition for the importance of setting reasonable limits on personalisation. Effectively addressing these challenges involves embracing transparent data practices, implementing robust consent mechanisms, and navigating the intricate path between customisation and ensuring user comfort, he explains.

Pitfalls and risks

In AI-driven ad personalization, companies face pitfalls such as data privacy concerns, bias in algorithms, over-reliance on AI, inaccurate predictions, and lack of transparency. To mitigate these risks effectively, companies should prioritise ethical data usage, ensuring compliance with privacy regulations and ethical standards, while also utilising diverse and representative data sets to train AI models.

Implementing human oversight and review processes can help monitor algorithms for fairness and accuracy, while transparent AI algorithms provide explanations for ad targeting decisions, fostering understanding and trust.

Adam Smart observes that even with the advent of AI, the pitfalls and risks of ad personalisation have not changed drastically.

“We still face the same challenges as before, but the increasing use of AI has reinforced them. Think of the risk of stereotyping and bias – AI algorithms may make assumptions based on user data, leading to discriminatory ad personalisation that could alienate certain demographic groups. And AI is also not immune to errors: AI models often rely on historical user data to make assumptions about preferences. This can lead to a static view of user preferences and may not account for changes in user behaviour or evolving preferences, or even to a misinterpretation of user behaviour, leading to inaccurate assumptions about preferences. For example, a one-time purchase might be wrongly interpreted as a long-term interest, resulting in misguided ad personalization,” he explains.

He advises companies to mitigate the risks associated with assumptions made by AI in ad personalization and build more responsible and effective advertising strategies.

While AI-driven ad personalisation is effective, it is not without drawbacks, states Hiren Shah. According to him, AI algorithms’ assumptions pose a serious concern, as they may result in erroneous user profiling and tailored content distribution.

“This can lead to mismatched advertisements, potentially frustrating or alienating consumers. To mitigate these risks, businesses should engage in ongoing monitoring and validation processes. Regularly updating algorithms based on user feedback, employing various data sets, and including human oversight can help assure the quality and relevancy of tailored information, minimising the potential pitfalls associated with AI assumptions,” Shah says.

Potential pitfalls in AI-driven ad personalisation include algorithmic bias, reinforcing stereotypes, and privacy infringement, says Dhaval Gupta.

To mitigate risks, he adds, organisations should regularly audit algorithms for bias, diversify training datasets, and implement transparency in AI decision-making. Adopting privacy-centric practices, obtaining explicit user consent, and providing opt-out options contribute to responsible personalization. Continuous monitoring, ethical guidelines, and collaboration with diverse teams help companies proactively address and rectify assumptions, ensuring AI in ad personalisation aligns with ethical standards and respects user privacy. Do remember, AI that has biases or is not able to maintain quality data will be trained in non-optimal ways. This can have a direct impact on the quality of your AI output as well. So not only is it good karma, it is also good quality,” Gupta says.

(Tomorrow, Part 2 of this report will discuss how companies can find the optimal balance between leveraging the scalability provided by generative AI tools and ensuring the implementation of validated, impactful personalization strategies in their advertising campaigns.)