Mehul Sharma and Daniel Lijo are students at Rajiv Gandhi National University of Law, Patiala (NLU Punjab) & National University of Advanced Legal Studies, Kochi (NUALS Kochi)
INTRODUCTION
The post dot-com era witnessed a sea change in advertising. Digital advertising in India grew 17 percent in 2024 to reach INR 700 billion, constituting 55 percent of total advertising revenues. Search and social media platforms accounted for 69 percent of India’s digital advertising revenues in 2024. In 2025, digital media revenues crossed INR 1 trillion for the first time. Followingly, the rise of ‘behavioural advertising’ came to dominate the digital epoch. It is known as the practice of targeting advertisements at individuals based on detailed profiles constructed from their online activity, browsing and search history, location data, purchase behaviour, device usage, and inferred interests, aggregated across websites, platforms, and time.
By relying on extensive collection of personal data, behavioural advertising raises concerns about an individual’s privacy. In this scenario, the adequacy of India’s Digital Personal Data Protection Act, 2023 (‘DPDPA/the Act’) merits analysis. The Act makes no mention or reference to profiling directed at adults. The only remnant of a protection lies in Section 9(3), which offers protection against targeted advertisement directed at children. This piece therefore, examines the shortcomings of the DPDPA in light of behavioural advertising, the inherent risks it carries and explores contextual advertising as a privacy-oriented alternative.
THE (MIS)BEHAVIOUR OF BEHAVIOURAL ADVERTISING UNDER THE DPDPA, 2023
A report by the German news website Netzpolitik.org revealed that over 6,50,000 categories were used by Xandr, a data management platform of Microsoft, to ‘sort’ its users to deliver targeted advertisements to them. These included labels such as “Moms who shop like crazy”, “Jewish”, “Period Trackers”, “Menopause”, “Sleep Disorders”, and “LGBT”, along with categories targeting individuals going through divorce, Black Lives Matter supporters, and anti-abortion groups. Such categorisation shows how behavioural advertising increasingly treats human identity, vulnerabilities, beliefs, and personal experiences as commercially exploitable data points. This concern is further aggravated by the inference-related problem, wherein platforms can infer deeply sensitive information about individuals even when such information was never consciously disclosed by the user. For instance, a user repeatedly visiting pages relating to fertility clinics, pregnancy forums, or baby products may be categorised as “Pregnant” even if the user never explicitly disclosed such information to the platform. The DPDPA fails to cater to such sensitive inferences, as the definition of “personal data” omits the explicit mention of inferences from it, unlike the Personal Data Protection Bill, 2019 (‘2019 Bill’) where personal data was defined to include “any inference drawn from such data for the purpose of profiling”. Due to this legal loophole, platforms might argue that the assignment of probabilistic advertising labels to users is not covered within the definition of “personal data” under the Act, since no directly identifiable information is processed and such labels are merely inferential classifications attached to pseudonymous identifiers.
Another lacuna under the Act in our case lies in the exclusion of publicly available personal data from its scope. Section 3(c)(ii) of the Act excludes personal data that a data principal has “made or caused to be made publicly available” meaning public social media posts, profiles, comments, etc, on open platforms are not granted protected status under the Act. This creates ambiguity and provides for a situation where a data fiduciary may carry out user profiling with no condition barring the qualification of such use. In India, the harms of behavioural advertising rear its head when read alongside Section 17(2)(a) of the DPDPA, which grants a blanket exemption to notified State instrumentalities from the Act entirely on broad grounds, including state security and public order. This creates an open corridor through which the Central Government may demand and receive behavioural profiles held by data fiduciaries about any data principal, with no restraint on how that data is subsequently processed or used. Section 17(4) further provides that, in respect of processing by the State or any instrumentality of the State, the right to erasure under Section 8(7) and the right to access information about processing under Section 12(2) and 12(3) do not apply.
A notable deficiency of the DPDPA is its failure to recognise the broader harms that may arise from the processing of personal data. For instance, a user who frequently watches fitness and weight-loss videos may be repeatedly shown advertisements for diet products or cosmetic procedures, which can deepen existing insecurities and affect their decisions. Such a practice may lead to psychological manipulation, one where vulnerabilities are capitalised. Beyond privacy concerns, behavioural advertising may actually produce a chilling effect on a user’s autonomy. Awareness that one’s searches, reading habits, and online activities are continuously tracked and profiled may discourage individuals from seeking information on sensitive subjects, shaping their behaviour through constant surveillance itself. Conversely, the DPDPA provides no category of inherently risky processing, no mandatory data protection impact assessments for behavioural profiling at any scale and it recognises no non-material harm as independently cognizable. The growing recognition of such harms is evident from the recent verdict by a Los Angeles jury on 25 March 2026, which found Meta and YouTube negligent for designing platforms addictive and harmful to young people, awarding USD 6 million in the first civil trial to hold major technology companies liable for harms associated with addictive platform design and resulting psychological harm. At a time when courts across jurisdictions are beginning to recognise psychological and other non-material harms arising from digital platforms, the very concept of “harm” does not find its existence in the DPDPA.
READING THE ROOM THROUGH CONTEXTUAL ADVERTISING
The initial draft of the American Privacy Rights Act defined contextual advertising as, “when an advertisement is displayed online based on the content of the webpage or online service on which the advertisement appears.” In its updated draft of the American Privacy Rights Act, the scope was widened and contextual advertising was defined as online advertisements that are not targeted based on an individual’s identity or behavioural profile, but are shown solely on the basis of the content of a webpage or online service, a user’s specific request for information or feedback, or the user’s presence within a broad geographic radius of at least 10 miles. Further, Centre for Democracy and Technology has explained that the ‘context’ in advertising would range differently when talking about different digital environments, i.e., in web (based only on what is currently present on the screen when a user visits a webpage), in mobile app (based only on the app or content currently being used, the user’s activity or searches, and not on other apps installed on the device), and in video or audio streaming service (based only on the content currently being watched or listened to, including immediately adjacent content, and not on the user’s broader watch history).
However, contextual advertising in the context of social media can be quite difficult to define since social media platforms rely extensively on the usage of personal information and behavioural profiling for targeting advertisements. The Center for Democracy & Technology notes that the context on social media sites can refer to the actual page, community, searched keywords, and content that are currently in view, but it cannot include the actual information about the person like his age or gender. Accordingly, contextual advertising on social media sites can mean placing ads into appropriate communities or content as well as placing them in response to a certain search query. For instance, when a user visits a website giving Wimbledon-related news updates or searches for stores selling tennis rackets and is shown advertisements for tennis rackets solely because of the content being viewed or the immediate search query, such advertisement can be considered contextual advertising. Similarly, advertisements for tennis shoes, coaching academies, or sports equipment shown while a user is viewing tennis-related posts, communities, hashtags, or videos on social media would also amount to contextual advertising. On the other hand, if the same advertisements are shown based on the user’s demographics, purchase history, watch history, page interactions, or other data collected over time, they would cease to be contextual advertising. This distinction is important for India, as it would help ensure that advertisements are treated as contextual only when they are based on the content being viewed or the immediate search query, and not on an individual’s past online activity.
However, not all advertisements described as contextual are necessarily based on context alone. Some platforms may combine contextual information with users’ personal data and past online activities to decide which advertisements to display. Therefore, an advertisement should qualify as purely contextual only when its selection is based on the content being viewed or the immediate search query, rather than the user’s behavioural profile.
CONTEXTUALISING POLICY TO PRACTICE
References to contextual advertising can be found across several state data protection laws in the US, where it is often expressly excluded from the scope of “targeted advertising”, including under the Section 6-1-1303(25)(b)(III) of the Colorado Privacy Act, Section 59.1-575 of the Virginia Consumer Data Protection Act, and Section 1(28)(B) of the Connecticut Data Privacy Act. Further, the Federal Trade Commission opines that “contextual advertising provides greater transparency than other forms of behavioural advertising, is more likely to be consistent with consumer expectations, and presents minimal privacy intrusion when weighed against the potential benefits to consumers”. Similarly, the United Kingdom’s (‘UK’) Information Commissioner’s Office in its guidance stated that contextual advertising more readily enables compliance with both Privacy and Electronic Communications Regulations requirements and UK GDPR obligations than other types of targeted advertising. When compared to behavioural advertising, the European Data Protection Board considers contextual advertising to be a less privacy intrusive alternative where data subjects are given an opportunity to exercise real choice.
Even in industrial practice, contextual advertising has proved to be commercially beneficial. The global contextual advertising market was valued at approximately USD 211.62 billion in 2024 and is projected to grow to USD 233.89 billion in 2025 and USD 345.24 billion by 2029, representing a compound annual growth rate of 10.2 percent during the forecast period. Notably, contextually relevant advertisements have also demonstrated stronger consumer responses: one neuroanalytical study found that such advertisements generated 43 percent greater neural engagement and were 2.2 times more memorable than contextually irrelevant advertisements, while also producing a statistically significant increase in purchase intent. Another study across 76 advertising campaigns across 8 different sectors suggested that even in the cases of skippable ads (which are usually the hardest ads to keep users engaged with), when placed contextually, deliver 70 percent more attention than average skippable ads. Post the introduction of Apple’s App Tracking Transparency framework in 2021 (which requires apps to obtain user permission before tracking them across apps and websites), many users refused tracking permissions. Due to this, platforms had to restrict their profiling based behavioral advertising and Facebook’s Audience Network ad transitioned to contextual targeting, eventually taking a revenue hit of USD 10 billion. In September 2025, Jio deployed Moment.AI, an AI-driven contextual advertisement tool which scans content libraries to detect more than 600 emotional and visual signals to optimize placement in video content, with early results indicating a 34 percent improvement in advertising effectiveness compared to conventional placements.
THE WAY FORWARD
Behavioural advertising has made our online activities and personal information valuable to advertisers, while the DPDPA leaves important gaps in dealing with profiling, inferences and the harms they may cause. Given its legal and practical benefits, India should therefore recalibrate its advertising ecosystem by treating contextual advertising as the normative baseline, as it allows advertisements to be displayed based on the content being viewed rather than an individual’s past behaviour. To implement this, DPDPA should introduce an explicit definition of profiling and inferred personal data, take into account psychological and non-pecuniary harm that may result from data processing. Moreover, the law must require the implementation of safety measures such as Data Protection Impact Assessments in the case of large-scale profiling practices. The Government could also introduce clear guidelines defining contextual advertising and distinguishing it from behavioural advertising, while allowing users to have a better alternative. Ultimately, the DPDPA must ensure that consumers are not stripped of their autonomy and corporations do not find reprieve in the absence of law or fault of its design.
***
