Hubbry Logo
Cross-device trackingCross-device trackingMain
Open search
Cross-device tracking
Community hub
Cross-device tracking
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Cross-device tracking
Cross-device tracking
Cross-device tracking
View on Wikipedia
from Wikipedia

Cross-device tracking is technology that enables the tracking of users across multiple devices such as smartphones, television sets, smart TVs, and personal computers.[1][2]

More specifically, cross-device tracking is a technique in which technology companies and advertisers deploy trackers, often in the form of unique identifiers, cookies, or even ultrasonic signals, to generate a profile of users across multiple devices, not simply one.[3] For example, one such form of this tracking uses audio beacons, or inaudible sounds, emitted by one device and recognized through the microphone of the other device.[3]

This form of tracking is used primarily by technology companies and advertisers who use this information to piece together a cohesive profile of the user.[3] These profiles inform and predict the type of advertisements the user receives.[3]

Background

[edit]

There are many ways in which online tracking has manifested itself. Historically, when companies wanted to track users' online behavior, they simply had users sign in to their website.[4] This is a form of deterministic cross-device tracking, in which the user's devices are associated with their account credentials, such as their email or username.[5] Consequently, while the user is logged in, the company can keep a running history of what sites the user has been to and which ads the user interacted with between computers and mobile devices.[5]

Eventually, cookies were deployed by advertisers, providing each user with a unique identifier in his or her browser so that the user's preferences can be monitored.[6] This unique identifier informs the placement of relevant, targeted ads the user may receive.[6] Cookies were also used by companies to improve the user experience, enabling users to pick up where they left off on websites.[7] However, as users began using multiple devices––up to around five––advertisers became confused as to how to track, manage, and consolidate this data across multiple devices as the cookie-based model suggested that each device––whether a phone, computer, or tablet––was a different person.[6]

Other technologies such as supercookies, which stay on computers long after the user deletes his or her cookies, and web beacons, which are unique images from a URL, are also used by trackers and advertisers to gain increased insight into users' behavior.[6] However, advertisers were still limited in that only one device was able to be tracked and associated with a user.[6]

Thus, cross-device tracking initially emerged as a means of generating a profile of users across multiple devices, not simply one.

One such tactic for cross-device tracking is called browser fingerprinting, and occurs when browsers, which are modifiable to the users' tastes, produce a unique signal that companies or advertisers can use to single out the user.[6] Browser fingerprinting has been a cause for concern because of its effectiveness and also since it does not allow for users to opt-out of the tracking.[6]

Another tactic used by Google is called AdID and works on smartphones in tandem with cookies on a user's computer to track behavior across devices.[4]

Now, cross-device tracking has evolved into a new, radical form of surveillance technology which enables users to be tracked across multiple devices, including smartphones, TVs, and personal computers through the use of audio beacons, or inaudible sound, emitted by one device and recognized through the microphone of the other device, usually a smartphone.[3] In addition, cross-device tracking may presage the future of the Internet of things (IoT), in which all types of devices––such as offices, cars, and homes––are seamlessly interconnected via the internet.[4]

Applications

[edit]

Studies have shown that 234 Android applications are eavesdropping on these ultrasonic channels without the user's awareness.[3]

Applications such as SilverPush, Shopkick, and Lisnr are part of an "ultrasonic side-channel" in which the app, often unbeknownst to the user, intercepts ultrasonic signals emitted from the user's environment, such as from a TV, to track which advertisements the user has heard and how long the person listened to them.[3]

  • SilverPush––the leading company using this technology––patented software enabling them to track TV ads based on audio stream above[3]
  • Shopkick, another popular application, gives discounts to users who shop at stores which emit these ultrasonic beacons, allowing them to create a profile of the user[3]
  • Lisnr uses a user's location data in tandem with ultrasonic beacons to give users coupons related to their activities[3]

Another study suggested that Apple, Google, and Bluetooth Special Interest groups need to do more to prevent cross-device tracking.

[8]

Privacy and surveillance concerns

[edit]

Ultrasonic tracking

[edit]

Humans interpret sound by picking up on different frequencies.[3] Given the variety of sound waves that exist, humans can only hear frequencies that are within a certain range––generally from 20 Hz to 20 kHz. By the age of 30, most humans cannot hear sounds above 18 kHz.[3]

Ultrasound, which is shorter wavelengths greater than or equal to 20 kHz, enables the rapid transmission of data necessary for cross-device tracking to occur.[3]

Another integral component of cross-device tracking is the usage of audio beacons. Audio beacons are beacons that are embedded into ultrasound, so they cannot be heard by humans.[3] These audio beacons are used to surreptitiously track a user's location and monitor online behavior by connecting with the microphone on another device without the user's awareness.[3]

In October 2015, the Center for Democracy and Technology submitted comments to the Federal Trade Commission (FTC) regarding cross-device tracking technology, specifically mentioning SilverPush.[9][10]

Audio "beacons" can be embedded into television advertisements. In a similar manner to radio beacons, these can be picked up by mobile apps.[11] This allows the behavior of users to be tracked, including which ads were seen by the user and how long they watched an ad before changing the channel.[12]

In March 2016, the FTC issued warning letters to 12 app developers using cross-device tracking in their apps.[13] The FTC warned these developers that they may be violating the FTC Act if they state or imply that their apps are not tracking television viewing habits when they in fact are.

Cross-device tracking has privacy implications and allows for more detailed tracking of users than traditional tracking methods. Data can be collected from multiple devices used by a single user and correlated to form a more accurate picture of the person being tracked.[12] Moreover, malicious actors may use variants of the technology to de-anonymize anonymity network users.[14]

Ultrasonic tracking technologies can pose massive threats to a user's privacy. There are four primary privacy concerns associated with this new form of tracking:

  • The first is media tracking: audio from the user's television may be detected by the microphone in the user's mobile device, allowing malicious actors to gain access to what the user is watching––particularly if it is salacious.[3] Advertisers can similarly gain insight into what a user typically watches.[3] In both scenarios, a user's real-world behavior is linked to their online identity and used for tracking.[3]
  • Another form of tracking permitted by ultrasonic tracking is cross-device tracking, which enables a user's profile to be connected across multiple devices based on proximity.[3] This form of tracking, in linking different devices, can help advertisers show more targeted ads or open individuals to attacks by malicious actors.[3]
  • Location tracking is yet another privacy concern.[3] Indeed, ultrasonic signals can convey location information via a location identifier, often placed in stores or businesses.[3]
  • Lastly, this new ultrasonic tracking poses a threat to users of Bitcoin and Tor because it de-anonymizes a user's information, since ultrasonic signals associate the user's mobile phone with the Bitcoin or Tor account.[3]

Panoptic surveillance and the commodification of users' digital identity

[edit]

From cookies to ultrasonic trackers, some argue that invasive forms of surveillance underscore how users are trapped in a digital panopticon, similar to the concept envisioned by Jeremy Bentham: a prison in which the prisoners were able to be seen at all times by guards but were unable to detect when, or even if, they were being watched at all, creating a sense of paranoia that drove prisoners to carefully police their own behavior.[15] Similarly, scholars have drawn parallels between Bentham's panopticon and today's pervasive use of internet tracking in that individuals lack awareness to the vast disparities of power that exist between themselves and the corporation to which they willingly give their data.[15] In essence, companies are able to gain access to consumers' activity when they use a company's services.[15] The usage of these services often is beneficial, which is why users agree to exchange personal information.[15] However, since users participate in this unequal environment, in which corporations hold most of the power and in which the user is obliged to accept the bad faith offers made by the corporations, users are operating in an environment that ultimately controls, shapes and molds them to think and behave in a certain way, depriving them of privacy.[15]

In direct response to the panoptic and invasive forms of tracking manifesting themselves within the digital realm, some have turned to sousveillance: a form of inverse surveillance in which users can record those who are surveilling them, thereby empowering themselves.[16] This form of counter surveillance, often used through small wearable recording devices, enables the subversion of corporate and government panoptic surveillance by holding those in power accountable and giving people a voice––a permanent video record––to push back against government abuses of power or malicious behavior that may go unchecked.[16]

The television, along with the remote control, is also argued to be conditioning humans into habitually repeating that which they enjoy without experiencing genuine surprise or even discomfort, a critique of the television similar to that of those made against information silos on social media sites today.[17] In essence, this technological development led to egocasting: a world in which people exert extreme amounts of control over what they watch and hear.[17] As a result, users deliberately avoid content they disagree with in any form––ideas, sounds, or images.[17] In turn, this siloing can drive political polarization and stoke tribalism.[17] Plus, companies like TiVO analyze how TV show watchers use their remote and DVR capability to skip over programming, such as advertisements––a privacy concern users may lack awareness of as well.[17]

Some scholars have even contended that in an age of increased surveillance, users now participate online through the active generation and curation of online images––a form of control.[3] In so doing, users can be seen as rejecting the shame associated with their private lives.[3] Other scholars note that surveillance is fundamentally dependent upon location in both physical and virtual environments.[18] This form of surveillance can be seen in travel websites which enable the user to share their vacation to a virtual audience.[18] The person's willingness to share their personal information online is validated by the audience, since the audience holds the user accountable and the user vicariously experiences pleasure through the audience.[18] Further, users' mobile data is increasingly being shared to third parties online, potentially underscoring the regulatory challenges inherent in protecting users' online privacy.[19]

In addition, scholars argue that users have the right to know the value of their personal data.[20] Increasingly, users' digital identity is becoming commodified through the selling and monetizing of their personal data for profit by large companies.[20] Unfortunately, many people appear to be unaware of the fact that their data holds monetary value that can potentially be used towards other products and services.[20] Thus, scholars are arguing for users' to have increased awareness and transparency into this process so that users can become empowered and informed consumers of data.[20]

Surveillance capitalism

[edit]

The increased usage of cross-device tracking by advertisers is indicative of the rise of a new era of data extraction and analysis as a form of profit, or surveillance capitalism, a term coined by Shoshana Zuboff.[21] This form of capitalism seeks to commodify private human experience to create behavioral futures markets, in which behavior is predicted and behavioral data is harvested from the user.[21] Zuboff suggests that this new era of surveillance capitalism eclipses Bentham's panopticon, becoming far more encroaching and invasive as, unlike a prison, there is no escape, and the thoughts, feelings, and actions of users are immediately extracted to be commodified and resold.[21] Thus, since cross-device tracking seeks to create a profile of a user across multiple devices, big tech companies, such as Google, could use this behavioral data to make predictions about the user's future behavior without the user's awareness.[21]

Scholars are beginning to discuss the possibility of quantifying the monetary value of users' personal data. Notably, the algorithms used to extract and mine user data are increasingly seen as business assets and thus protected via trade secrets.[20] Indeed, the usage of free online services, such as public Wi-Fi, often comes at the unknown cost to the user of being tracked and profiled by the company providing the service.[20] In essence, a transaction is occurring: users' personal data is being exchanged for access to a free service.[20] Increasingly, scholars are advocating for users' right to understand the fundamental value of their personal data more intimately so as to be more savvy, informed consumers who have the ability to protect the privacy of their online information and not be manipulated into unwittingly giving away personal information.[20]

Health and wellness applications

[edit]

In addition, health and wellness applications also have a dearth of privacy protections as well: a study found that many health apps lacked encryption and that regulators should enforce stronger data privacy protections.[22] The study stated that of the 79 apps they tested, none of the applications locally encrypted the users' personal information and 89% of the applications pushed the data online.[22] The lack of adequate privacy and security measures surrounding users' personal medical data on mobile applications underscores the lessening degree to which users can trust mobile app developers to safeguard their personal information online.[22] While mobile application developers continue to confront privacy and security concerns, users are increasingly looking to ways to visualize their data through wearable devices and applications that track their workout and exercise routines.[23] Indeed, researchers discovered that these self-tracking devices play a role as a tool, a toy, and a tutor in users' lives.[24] In the tool role, the self-tracking device functions as a mechanism to help the user in some capacity, often to achieve personal health goals.[24] The toy role underscores how some self-tracking users see it as a fun game, particularly with regard to rewards and viewing the visualized data.[24] Lastly, the tutor role reflects how users gain insights from and motivation about their activity from the apps themselves.[24] Other scholars have characterized self-tracking as performing for the system, or controlling what is (or isn't) recorded, performing for the self, tracking themselves to gain insight into their behavior, and performing for other people, or the importance of how other people viewed the person being tracked, as well as the control the person being tracked had over their data and thus how they are perceived.[25]

Cookies, flash cookies, and web beacons

[edit]

Additionally, privacy concerns surround cookies, flash cookies, and web beacons on websites today.[25] Ultimately, five main concerns surround the usage of cookies, flash cookies, and web beacons, according to a study:[7]

  • Firstly, the authors note that users lack anonymity online, with cookies using unique identifiers and flash cookies enabling recognition of website visits[7]
  • Another concern the authors note is unintended uses of cookies, since cookies were initially designed to benefit the user's experience and engagement online, but have since morphed into a business run by advertisers in which personal data is sold for profit[7]
  • Users are likely unaware of how their personal information is being used, reflecting the surreptitious nature of data collection[7]
  • Some cookies trespass into the web users' own resources and are downloaded to the user's computer often without the user's awareness[7]
  • Lastly, the authors note that the threat of cookie sharing underscores how web users' personal information can become combined with other data from websites and even a social security number to create a more cohesive picture of the user[7]

Data capitalism

[edit]

Other scholars have defined a similarly extractive and destructive phenomenon called data capitalism.[26] Data capitalism is an economic system enabling the redistribution of power towards those who have access to the information––namely, big corporations.[26] There are three fundamental theories of how large companies engage users in virtual communities, reflecting the power of data capitalism on users today:

  • The free and open network: in making products free, large companies make their products more accessible to a larger audience from which they can extract valuable data in exchange.[26]
  • The connection between people and machines: data capitalism promotes a connection between people and machines which is derived from the user's relationship to the technology itself.[26] Increasingly, tracking and surveillance technology is profiling users and learning their preferences, users become more comfortable with their devices and a self-fulfilling prophecy continues.[26]
  • The value placed on data: new information asymmetries are proliferating that exacerbate inequality of information and allow only the most powerful access to most people's data.[26] Increasingly, a scholar suggests that the lack of transparency over users' data reflects the tension between privacy and community online.[26]

Solutions

[edit]

Scholars are convinced the current notice-and-consent model for privacy policies is fundamentally flawed because it assumes users intuitively understand all of the facts in a privacy policy, which is often not the case.[27] Instead, scholars emphasize the imperative role of creating a culture in which privacy becomes a social norm.[27] In effect, users of online technologies should identify the social activities they use on the internet and start questioning websites' governing norms as a natural outgrowth of their web browsing.[27] In effect, these norms need to prevent websites from collecting and sharing users' personal information.[27] In addition, starting with a user's personal values and seeing how these values correlate with online norms may be another way to assess whether or not privacy norms are being violated in odd cases.[27] Ultimately, scholars believe these privacy norms are vital to protecting both individuals and social institutions.[27]

[edit]

While the United States lacks extensive privacy rights, the Fourth Amendment provides some privacy protections.[7] The Fourth Amendment states that "the right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated", suggesting that while individuals are protected from all levels of the government, they are not legally protected from private companies or individuals with malicious intent.[7]

There are large implications for this technology within the legal field. Legally, The Federal Trade Commission has a responsibility to prevent deceptive practices by technology companies, such as those that could lead to consumer injury.[28] The FTC has made efforts to prevent invasive web tracking, tracking in physical space, malware, insecure and poorly designed services, and the use of deception to engage in surveillance.[28] For instance, in the realm of invasive web tracking, the FTC has brought lawsuits against companies who engage in history sniffing, a technique that enables companies to ascertain which links a user clicked on based on the color of the link.[28] Concerning tracking in physical space, the FTC has also cracked down on Nomi, a company that scans the MAC addresses of customers' phones in stores.[28] MAC addresses function as a unique identifier, enabling the connection to wireless networks.[28] In the case of malware, the FTC has placed pressure on companies such as CyberSpy, a self-proclaimed email attachment company that claimed to secretly record users' key presses.[28] The FTC has also cracked down on companies like Compete, a browser toolbar, because it decrypted users' personal information on the internet, putting users at risk.[28] Lastly, in cases during which deception is used to engage in surveillance, the FTC has investigated private investigators, who surveil individuals on another person's behalf.[28] In addition, audio beacon technology, used by an application called Silverpush, could violate the FTC's policies because users were not made aware as to when the ultrasonic signals were being recorded.[28]

Another scholar believes that the convergence between lived experience and online technology is creating a term called mixed reality, in which people and things are replaced with virtual experiences.[29] Mixed Reality technologies can pose legal challenges in that laws which govern the online world will also extend to the real world.[29] In addition, data tagging––often through GPS, location-based services, or even near-field communication (NFC)––is the new technology at the heart of mixed reality, since people's data is determined in part by their location.[29] Near-field communication enables devices to transmit data to each other with a certain range.[29] Virtual reality can become a privacy issue because it attempts to immerse users into the virtual environment by recording a user's every sensation.[29] In turn, mixed reality's amalgamation with daily tasks suggest that it will be implicated in numerous legal issues ranging from copyright law to intellectual property law.[29] Customers are also being denied a voice in contracts, since only corporations set the rules by which individuals' private information is mined and extracted.[29] The solution to these issues, according to scholars, are opt-in controls to police users' privacy that enable balance to be restored to the law, particularly as it stands regarding contracts.[29]

Ethically, Zuboff points to the extraction, commodification, and analysis of private human experiences as well as increased surveillance––which is sometimes hidden––in everyday life as violating users' rights to privacy.[21] The usage of surreptitious methods, in which the user is unaware of the extent to which he or she is being tracked, brings tracking mechanisms––such as cookies, flash cookies, and web beacons––into the ethical realm as well since users are not being informed of this tracking perhaps as often as they should.[7]

See also

[edit]

References

[edit]
[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Cross-device tracking

View on Grokipedia
from Grokipedia
Cross-device tracking is a set of technologies and techniques used primarily in digital advertising and to detect, link, and attribute user behaviors and identities across multiple devices—such as smartphones, tablets, laptops, and smart TVs—owned or operated by the same , thereby enabling the construction of more complete behavioral profiles than single-device tracking alone permits. These methods address the fragmentation caused by users switching devices, allowing advertisers to measure cross-session interactions, such as a mobile ad exposure leading to a desktop conversion, which enhances attribution accuracy and campaign optimization. The core approaches divide into deterministic tracking, which relies on explicit, verified identifiers like logged-in email addresses, device IDs, or hashed to achieve high-confidence linkages, and probabilistic tracking, which employs statistical algorithms analyzing patterns in IP addresses, timestamps, geolocation signals, and browsing behaviors to infer probable device ownership with varying degrees of accuracy. Deterministic methods offer precision but depend on user authentication, while probabilistic ones scale broadly but risk false positives, such as misattributing shared household devices. Despite enabling measurable improvements in efficiency and detection, cross-device tracking has sparked substantial controversies, as it often operates opaquely, aggregating sensitive without clear user notification or granular control, potentially enabling pervasive and re-identification even in anonymized datasets. Regulatory responses, including the European Union's GDPR and California's CCPA, impose requirements, minimization obligations, and opt-out mechanisms that challenge implementation, particularly for probabilistic techniques reliant on inferred signals, while highlighting tensions between commercial utility and individual autonomy over flows.

Definition and Fundamentals

Core Principles

Cross-device tracking involves identifying and associating a user's activities across multiple devices—such as smartphones, tablets, desktops, and connected televisions—to form a cohesive profile of their and preferences. This process links separate sessions originating from different devices by leveraging shared signals, including user logins, IP addresses, and behavioral patterns, thereby aggregating fragmented into a holistic view that transcends individual device boundaries. Such linkage addresses the inherent silos in collection, where user interactions are often captured in isolation, enabling advertisers and analysts to reconstruct complete pathways without relying on incomplete, device-specific records. In contrast to single-device tracking, which limits observation to behaviors confined to one apparatus and thus underrepresents user actions, cross-device tracking accounts for the prevalent multi-device habits of consumers navigating digital ecosystems. Empirical reveals that 98% of Americans switch between devices within a single day, while the typical consumer possesses about 3.64 connected devices and 80% routinely alternate among them during online activities. This multi-device fragmentation necessitates cross-device methods to accurately connect user identities and infer continuities in behavior, such as research initiating on a culminating in a purchase via desktop, which single-device approaches would misattribute or overlook. At its core, cross-device tracking is empirically grounded in the documented dispersion of user interactions across devices, which generates disjointed datasets absent deliberate unification, thereby hindering reliable attribution of outcomes like ad-driven conversions to upstream stimuli. By establishing verifiable connections through observable signals, it permits causal mapping of user journeys—distinguishing true multi-session progressions from spurious device-independent events—and supports for profiling without presuming device autonomy mirrors user continuity. This principle-driven aggregation enhances the fidelity of behavioral insights, rooted in the reality that unlinked yields incomplete causal chains in digital analytics.

Deterministic and Probabilistic Approaches

Deterministic approaches to cross-device tracking establish linkages between devices using explicit, unique identifiers provided by users, such as addresses, phone numbers, or hashed credentials, enabling precise one-to-one matching when users are across platforms. These methods achieve high accuracy rates, typically exceeding 90%, because they rely on verifiable, exact matches rather than inference, making them particularly effective for signed-in users on services like Google's ecosystem where account data directly ties activities across devices. However, deterministic tracking is limited to scenarios with consistent user authentication, resulting in lower coverage for anonymous or inconsistently logged-in sessions. In contrast, probabilistic approaches infer device associations through statistical modeling of behavioral patterns and signals, including temporal overlaps in activity, geolocation similarities, correlations, and device usage heuristics, without requiring direct identifiers. This method applies to both authenticated and anonymous users, offering broader but with reduced precision, as match rates generally fall between 60% and 80% due to the inherent uncertainties in probabilistic , which can lead to false positives from coincidental similarities. Patents and technical analyses from the early , including those evaluating features for probabilistic linking, highlight its reliance on aggregated signals to approximate user identity graphs, though varies with and model sophistication. The core trade-off between these approaches lies in accuracy versus coverage: deterministic methods prioritize reliability for high-confidence attribution, often underestimating total cross-device activity by excluding non-logged sessions, while probabilistic techniques expand reach at the cost of potential errors, with studies indicating they can enhance overall attribution resolution by 20-30% in hybrid systems by filling gaps in deterministic . Real-world depends on signal richness; for instance, combining overlapping deterministic anchors with probabilistic modeling mitigates error rates, but pure probabilistic systems remain susceptible to from shared or demographic proxies.

Historical Evolution

Pre-2010 Foundations

HTTP cookies, invented in 1994 by at Communications, served as the foundational mechanism for maintaining state across stateless HTTP requests, enabling basic session continuity such as remembering user preferences or contents on websites. Initially designed to address the limitations of the web's , where each page load was independent, cookies allowed servers to store small data packets on client browsers, which were then sent back with subsequent requests. This primitive tracking capability laid the groundwork for identifying returning visitors, primarily within single-browser sessions on desktop computers dominant in the 1990s. Complementing cookies, web servers employed logging from the mid-1990s onward to delineate user sessions and approximate continuity, especially for non-cookie-enabled browsers or aggregate , by correlating request timestamps and source IPs despite challenges like dynamic addressing. Ad networks, such as founded in 1996, integrated these tools to track impressions and clicks across affiliated sites, using third-party cookies for cross-site user profiling that extended to rudimentary cross-browser linkage via shared identifiers. 's methods emphasized over device mobility, focusing on desktop environments where users typically operated within one primary browser instance. By the early 2000s, Adobe Flash's Local Shared Objects (LSOs), introduced around Flash MX in 2002, provided enhanced persistence by storing data outside browser silos, accessible across multiple browsers on the same device and resistant to standard deletions. This enabled advertisers to maintain tracking continuity amid users switching browsers, as LSOs operated at the Flash Player level rather than per-browser storage. Concurrently, the rapid adoption—rising from under 5% of U.S. households in 2000 to over 20% by 2003—supported longer, more data-intensive sessions, amplifying the utility of these for reliable user linkage without mobile fragmentation. Early deterministic elements emerged through email-based logins on portals, where ad networks matched hashed credentials to unify profiles across visits, predating widespread multi-device use but hinting at identity persistence.

2010s Expansion and Industry Adoption

The rapid proliferation of smartphones in the early , amid and Android platform fragmentation, spurred demand for cross-device tracking to link user behaviors across as mobile data traffic surged 26-fold globally from 2010 to 2015, with nearly one mobile-connected device per capita by the latter year. This multi-device usage pattern, fueled by ad tech investments, highlighted limitations in single-device attribution, prompting probabilistic modeling to infer connections via behavioral patterns and when deterministic identifiers like device IDs proved insufficient due to ecosystem silos. Mobile attribution firms such as Adjust, which emerged during 2010–2012 amid the app economy boom, integrated probabilistic approaches alongside deterministic methods to enable cross-device linkages, allowing advertisers to attribute installs and events across fragmented platforms without relying solely on exact matches. Similarly, advanced cross-device capabilities by the mid-decade, leveraging its user login data to construct graphs that bridged devices, as evidenced in analyses noting its dominance in probabilistic reach compared to competitors. The U.S. Federal Trade Commission (FTC) hosted a workshop on November 16, 2015, to examine cross-device tracking's privacy implications, raising awareness of consumer surveillance risks while implicitly validating its analytical utility for businesses through discussions of deterministic and probabilistic techniques. Concurrently, data brokers expanded device graphs aggregating billions of data elements on U.S. consumers, enabling probabilistic mappings that supported ad targeting and analytics despite transparency concerns outlined in the FTC's 2014 report on the industry. By 2016, Facebook's enhanced cross-device reports demonstrated measurable attribution gains, with integration revealing up to 45% additional impact from multi-device paths and mobile ads driving over half of campaign conversions previously unlinked in single-device models. These milestones underscored market-driven innovation, as firms prioritized probabilistic enhancements to capture conversion paths missed by siloed tracking, reflecting ad tech's emphasis on comprehensive user graphs over fragmented device views.

Post-2020 Regulatory Shifts

Apple's App Tracking Transparency framework, implemented in 14.5 on April 26, 2021, mandated explicit user consent for apps to access the (IDFA), curtailing deterministic cross-device linkage reliant on this persistent identifier. Opt-in rates for tracking permission averaged below 30% across major apps, with sectors like reporting as low as 24%, resulting in a 55-80% effective reduction in signal availability for iOS-based cross-device graphs and prompting a pivot to probabilistic inference models that aggregate behavioral patterns without direct identifiers. This shift diminished revenue for ad-dependent platforms by 10-30% in affected ecosystems, accelerating investments in privacy-compliant alternatives like contextual and cohort-based targeting. Google's progression toward deprecating third-party cookies in Chrome, announced amid mounting regulatory scrutiny from bodies like the UK's , involved expanded trials starting in early 2023, with phased restrictions reaching 1% of users by Q1 2024 and full rollout delayed to Q1 2025 pending final approvals. The within Sandbox enables server-side classification of user interests into temporary cohorts, supporting aggregated cross-device ad relevance without exposing individual histories, which trials indicated could preserve 80-90% of auction value in controlled environments despite cookie loss. This adaptation countered fragmentation in browser-based tracking, fostering reliance on first-party data stores and enhanced signals for multi-device user modeling. By 2023-2025, 4's integration of machine learning-driven conversion modeling compensated for signal loss under these regimes, using anonymized aggregates and predictive algorithms to link cross-device paths, yielding data-driven attribution accuracies comparable to pre-restriction baselines in enterprise deployments. Industry analyses reported sustained 10-20% lifts in cross-platform attribution completeness via these modeled approaches, even as deterministic inputs declined, underscoring regulatory incentives for AI-augmented probabilistic systems over outright abandonment of cross-device capabilities.

Technical Mechanisms

Identifiers and Data Linkage

Deterministic identifiers form the foundation of precise cross-device linkage by enabling exact matches between user activities on different devices. These include HTTP cookies for browser-based sessions, though primarily effective within the same device until synced via logins; advertising IDs such as Apple's IDFA (available prior to the 2021 App Tracking Transparency framework, which required user consent for tracking); and hashed personal identifiers like addresses or phone numbers to facilitate privacy-preserving matching without exposing raw data. Data linkage in deterministic approaches relies on direct equivalence of these identifiers across devices, often triggered by user events like logins that propagate IDs through hashed formats compliant with standards. For instance, email hashing—transforming emails via cryptographic functions like SHA-256 before comparison—allows platforms to link profiles without revealing personal information, as standardized in solutions like The Trade Desk's UID 2.0 framework endorsed by industry bodies. Probabilistic linkage complements this by fusing weaker signals for cases lacking exact IDs, employing statistical models to correlate IP addresses (indicating shared networks like households), user-agent strings (detailing browser and OS versions), and behavioral timestamps (aligning session times across devices) to estimate user continuity with tunable confidence thresholds. Device graphs emerge from this linkage process as scalable data structures mapping relationships between devices, users, and households, built by aggregating probabilistic signals over time to represent clusters like family-shared IPs or overlapping usage patterns. Engineering considerations prioritize signal reliability—IP for coarse geographic/household ties, user-agents for device fingerprint consistency, and timestamps for temporal proximity—to achieve linkage scalability across billions of events, though accuracy varies with data volume and noise from VPNs or shared networks. Industry guidelines, such as those from IAB Tech Lab on identity solutions, emphasize standardized hashing and signal validation to ensure robust, privacy-compliant fusion without over-reliance on any single input.

Device Graphs and Fingerprinting

Device graphs represent networked models that probabilistically or deterministically link multiple devices to a single user profile by aggregating shared attributes such as email addresses, IP overlaps, login events, or behavioral patterns observed across sessions. These graphs function as inference engines, constructing edges between nodes (devices) based on weighted probabilities derived from data linkages, enabling cross-device identity resolution without relying solely on cookies. For instance, Oracle's Identity Graph, which evolved from its acquisition of BlueKai, encompasses connections for over 220 million U.S. adults as of the early 2020s, facilitating linkage across disparate touchpoints by prioritizing high-confidence matches while estimating uncertainty in lower-confidence ones. Such models incorporate probabilistic techniques to refine linkages iteratively, akin to Bayesian updating, where successive signals adjust the posterior probability of a match, thereby mitigating false positives from noisy data like transient IP addresses. Browser and device fingerprinting complements device graphs by generating quasi-unique identifiers from passive signals inherent to a user's setup, bypassing traditional deprecation. Techniques include , which renders hidden canvases to capture variations in graphics processing (e.g., differences), font enumeration to detect installed typefaces via measured text metrics, and hardware-derived hashes from CPU details, screen resolution, or sensor . These signals are hashed into stable fingerprints that feed into graph models for probabilistic stitching, allowing of user continuity across sessions or devices even when direct identifiers are absent. Empirical studies indicate fingerprints achieve high uniqueness in uncontrolled environments—often distinguishing over 99% of users—but accuracy degrades substantially under interventions, such as browser extensions that inject noise into canvas renders or randomize font lists, reducing linkage precision to levels where matches become unreliable for sparse sets. Despite refinements, device graphs and fingerprinting exhibit inherent accuracy limits rooted in empirical variability and causal mismatches. Probabilistic linking reduces false positives by thresholding low-probability edges, yet over-reliance on aggregate signals can propagate errors in multi-user households, where shared access confounds individual attribution and ignores user-specific agency in device usage patterns. For example, Bayesian-inspired models estimate false linkage rates by modeling error distributions, but real-world deployment often underperforms due to unobservable confounders like VPN-induced IP shifts or evolving hardware, capping reliable coverage below universal claims and necessitating hybrid validation against logged identifiers where available. These constraints underscore that while graphs enable broad inference, their outputs remain probabilistic approximations rather than deterministic truths, with error rates rising in privacy-enhanced or heterogeneous environments.

Ultrasonic and Cross-Modal Tracking

Ultrasonic tracking utilizes inaudible high-frequency audio signals, generally between 18 and 22 kHz, emitted from speakers in media such as television advertisements, web pages, or retail environments. Mobile applications equipped with microphone access detect these signals, decode embedded identifiers, and facilitate the linkage of user sessions across devices or applications by associating the beacon data with existing tracking profiles. This method emerged in the mid-2010s, with patents such as US9024998B2 (filed May 2, 2013) describing ultrasonic beacons for device pairing in settings like videoconferences, and US20150215668A1 outlining audio streams mixed into media for cross-device targeting. Cross-modal variants combine ultrasonic audio with complementary signals, such as (BLE) advertisements or beacons, to corroborate proximity and refine device handoffs in scenarios like in-store navigation or multi-screen interactions. These hybrids aim to mitigate ultrasonic signal over distance but remain niche, constrained by the energy demands of perpetual sampling, which accelerates battery depletion on resource-limited devices. Implementation faces platform-specific hurdles, including iOS restrictions on background audio capture tightened since (2017), which require foreground app activity or explicit user authorization for use, effectively curbing opportunistic deployment. Ultrasonic signals are not inherently undetectable; tools like spectrum analyzers or apps such as SoniControl can identify and filter them by scanning for anomalous frequencies above the human audible range. concerns are thus mitigated by dependency on user-granted permissions and active app listening, with no evidence of zero-consent exploitation in standard consumer apps, leading to restrained adoption relative to more efficient alternatives like probabilistic fingerprinting.

Applications and Innovations

Advertising Attribution

Cross-device tracking facilitates advertising attribution by linking user interactions with ads on one device to subsequent conversions on another, such as an initial ad exposure via leading to a desktop purchase. This approach addresses the fragmentation in single-device measurement, where up to 70% of conversions may involve multiple devices in the . For instance, platforms like provide cross-device conversion reports that quantify assists, such as a mobile click contributing to two additional non-mobile conversions, enabling precise credit assignment across touchpoints. In ad ecosystems, this tracking enhances return on ad spend (ROAS) by unifying customer journeys, allowing marketers to calculate true attribution beyond device silos. Empirical data indicates that cross-device insights can lower by 30-50% through better-targeted follow-up ads, such as retargeting cart additions from mobile on desktop. Attribution models incorporating these linkages, as in 4's multi-touch enhancements, reveal hidden efficiencies, with studies showing improved measurement of full-funnel impacts. Innovations in cross-device auctions, such as Google's journey-aware bidding introduced in 2025, optimize real-time bids by factoring in multi-device paths, prioritizing high-value users regardless of screen. This reduces ad waste by focusing spend on verifiable cross-device signals, enabling smaller advertisers to achieve competitive ROAS without relying on broad, inefficient targeting. Overall, such mechanisms have demonstrated uplift in conversion accuracy, with platform-level analyses reporting 19-20% improvements in attributed outcomes from integrated tracking.

Analytics and Personalization

Cross-device tracking enables analytics platforms to construct unified user profiles by linking behaviors across devices, facilitating a holistic view of user interactions independent of advertising attribution. In 4 (GA4), the User-ID feature associates custom identifiers—typically from user logins—with sessions, allowing measurement of activity across multiple devices and platforms to reveal complete behavioral paths. This capability, enhanced through GA4's identity resolution methods including User-ID and Signals, supports accurate tracking of multi-device journeys, with implementation guides updated as recently as 2025 emphasizing its role in unifying fragmented data for non-advertising insights. Such unification informs strategies by aggregating user data into persistent profiles, enabling services to deliver tailored content and interfaces that span devices. For instance, leverages account-based cross-device to maintain consistent recommendation algorithms, where over 80% of viewed content derives from personalized suggestions informed by aggregated viewing history across smartphones, tablets, and televisions. This approach contributes to 's reported customer retention rate exceeding 98% in early 2025, as seamless multi-device experiences reduce friction in content discovery and consumption. Similarly, Amazon employs logged-in user graphs to synchronize shopping carts and preferences across devices, relying on voluntary to ensure continuity in browsing and purchasing sessions without probabilistic inferences. Empirical gains from these practices include improved user engagement metrics, as cross-device profiling allows platforms to anticipate needs based on verified, consented linkages rather than device silos. Studies and platform reports indicate that such enhances satisfaction by minimizing redundant inputs and aligning interfaces with evolving user contexts, though benefits accrue primarily to authenticated users who opt into account . This contrasts with anonymous tracking limitations, underscoring the value of explicit user identifiers for reliable, non-intrusive profiling in and service delivery.

Fraud Detection and Security

Cross-device tracking enhances detection by enabling the identification of anomalous patterns across user sessions and devices, such as those generated by bot farms or synthetic identities. Probabilistic device graphs, which probabilistically link user identifiers like , IP addresses, and browser fingerprints across devices, allow financial institutions to flag inconsistencies, including rapid succession of logins from mismatched device profiles or geolocations atypical of legitimate . This approach uncovers coordinated rings by mapping relationships between accounts, transactions, and devices, as demonstrated in graph-based models applied in banking operations. In account security, cross-device tracking prevents unauthorized access by associating ongoing sessions with verified device histories, triggering alerts for deviations like attempts from unfamiliar hardware. For instance, when a new device initiates a , systems it against prior usage patterns and may require via linked channels, thereby verifying legitimate ownership and reducing risks of or takeover. Financial institutions utilize this to monitor for , where fraudsters exploit stolen credentials across platforms. Payment processors like integrate device fingerprinting into risk models that evolve with transaction data, identifying infrastructure associated with fraud attempts through unique device signatures derived from hardware, software, and behavioral signals. This contributes to real-time threat mitigation by distinguishing genuine cross-device user continuity from malicious emulation.

Benefits and Economic Value

Marketing Efficiency and ROI

Cross-device tracking enhances marketing (ROI) by improving attribution accuracy, as users often initiate interactions on one device and convert on another, leading to under-attribution without linkage. Empirical data indicates that 30-50% of conversions involve multiple devices, enabling marketers to capture previously missed revenue streams through unified tracking. For instance, one implementation reported a 15% increase in measured conversion rates and a 25% reduction in ad spend after adopting cross-device methods, demonstrating quantifiable lifts in efficiency. In the context of global digital expenditures exceeding $740 billion in , cross-device capabilities minimize waste by facilitating precise targeting and frequency capping across devices, preventing redundant ad exposures to the same users. This precision counters under-attribution biases inherent in siloed device data, allowing reallocations toward high-performing channels based on holistic user journeys rather than fragmented metrics. Studies attribute such optimizations to direct ROI gains, as accurate cross-device insights reveal true contributions to conversions, reducing overall campaign inefficiencies. By democratizing access to comprehensive user data, cross-device tracking levels the playing field for small and medium-sized enterprises (SMEs), which otherwise struggle against larger firms with proprietary datasets. Smaller businesses have cited the need for these methodologies to compete effectively in personalized , fostering market without relying on monopolistic structures. This competitive dynamic underscores efficiency as an emergent market outcome, where empirical tracking advantages drive resource optimization amid rising ad volumes.

User-Centric Improvements

Cross-device tracking enables seamless continuity of user sessions and activities, minimizing disruptions in multi-device environments. Apple's Handoff feature, for example, permits initiating tasks like web browsing or document editing on an and resuming them instantaneously on a Mac, leveraging proximity-based detection and synchronization to preserve context without manual intervention. This reduces and login friction, fostering habitual cross-device usage as users experience uninterrupted workflows, even under Apple's restrictions on third-party trackers. By aggregating behavioral signals across devices, tracking supports more precise of interfaces and content feeds, causally linked to elevated user engagement through rather than volume. Empirical analyses of implementations, including those informed by cross-device data, report engagement improvements of 34% via tailored recommendations that align with observed patterns, as validated in A/B-controlled experiments. Such gains arise from users encountering fewer irrelevant interruptions, thereby extending session durations and interaction depth in platforms like streaming services or apps. Users opting into tracking-enabled features demonstrate higher retention rates, reflecting perceived utility in continuity and customization over default . Studies on voluntary self-tracking reveal preferences for automated monitoring that delivers actionable insights, correlating with sustained participation and satisfaction absent in non-opted scenarios. This pattern holds in cross-device contexts, where explicit aligns with first-party use for enhancements like synchronized histories, yielding retention lifts tied to experiential value rather than imposed defaults.

Broader Market Innovations

Cross-device tracking has accelerated advancements in for predicting consumer behavior by supplying interconnected datasets across devices, which train models to infer and preferences with greater . This capability underpins sophisticated ad tech platforms, where aggregated cross-device signals enable probabilistic modeling of user trajectories, enhancing at scale. The digital advertising , bolstered by such innovations, saw global spending exceed $600 billion in 2024, reflecting the economic scale of these AI-driven developments. Data brokers exemplify this aggregation's broader utility, compiling cross-device profiles into actionable B2B intelligence for sectors beyond consumer , including and . Firms like process vast datasets to generate predictive insights, such as demographic overlays and behavioral clusters, which inform enterprise and stimulate targeted efficiencies. The industry, valued at $303 billion in 2024, channels these capabilities into economic multipliers, supporting GDP contributions through refined market allocations akin to offline data practices like compilations. In , cross-device tracking has catalyzed post-2010s expansion by enabling consistent user identification and attribution across touchpoints, fostering seamless conversions from mobile browsing to desktop purchases. This underpinned a shift where 's share of retail rose from under 5% in 2010 to 18% by 2020, with global volumes projected at $6.3 trillion in 2024, driven by tracking-enabled that parallels offline analogs such as catalog targeting or card profiling, scaled without demonstrated net societal detriment beyond intensified data volumes.

Criticisms and Risks

Privacy Erosion Claims

Critics of cross-device tracking argue that it enables the aggregation of user data across devices to construct detailed behavioral dossiers, potentially eroding by facilitating persistent without explicit consent. However, mechanisms like rights under regulations such as the (CCPA), effective since 2018 and expanded in 2023 via the , allow consumers to limit data sharing, with businesses required to honor requests including Global Privacy Control signals. Enforcement actions by the California Privacy Protection Agency in 2023 and 2024 demonstrate the efficacy of these opt-outs, as companies faced penalties for non-compliance, thereby constraining the scope of unrestricted aggregation. Empirical evidence does not show widespread real-world harms such as spikes in directly linked to cross-device tracking; instead, surges in recent years have been attributed to factors like AI-generated synthetic identities, data sales, and general data breaches rather than tracking linkages. Studies on risks from digital tracking, including cross-device methods, highlight potential vulnerabilities but lack documentation of causal increases in identity theft or other tangible harms beyond theoretical risks. Data breaches, often cited in privacy erosion discussions, predominantly stem from hacking incidents, unauthorized internal disclosures, and poor storage practices like misconfigured or inadequate , not from the act of linking devices themselves. This distinction underscores that while aggregation raises concerns, the primary causal vectors for data exposure involve storage and access failures rather than tracking . From an economic perspective, cross-device tracking operates within a value exchange model where users receive free or subsidized services—such as personalized recommendations and ad-supported content—in return for contributions, akin to historical bartering systems where yields mutual benefits without net loss for informed participants. This framework, analyzed in economic assessments of digital markets, reveals that perceived erosion often overlooks the consumer surplus generated, as platforms monetize to sustain zero-price offerings that would otherwise require direct payment.

Surveillance Capitalism Critiques

Shoshana Zuboff's concept of surveillance capitalism posits that digital platforms unilaterally claim behavioral as a surplus resource, enabling the prediction and modification of user actions through opaque algorithms, thereby establishing an asymmetric power dynamic akin to instrumentarianism that undermines individual and democratic processes. This framework, articulated in her 2019 book, frames cross-device tracking as a cornerstone of this system, where aggregated across devices facilitates "reality engineering" without user consent, drawing parallels to a digital . Left-leaning critiques, including those from , acknowledge continuities with historical capitalist enclosures but argue Zuboff's emphasis on novelty overlooks entrenched power structures in and . Such portrayals overstate the monopoly on behavioral modification by disregarding market voluntarism and competitive pressures in digital advertising, where firms vie for user attention and data yields efficiencies rather than totalitarian control. Empirical analyses indicate no pervasive causal evidence of panopticon-like dominance; instead, targeted advertising, bolstered by cross-device tracking, mirrors traditional marketing's persuasive tactics but with measurable improvements in return on investment through precision, sustaining free services users voluntarily adopt. User agency manifests in widespread adoption of countermeasures, with approximately 31.5% of global internet users employing ad blockers at least occasionally as of early 2024, enabling opt-outs from tracking ecosystems. Privacy-oriented alternatives further refute claims of inescapable monopolization, as evidenced by DuckDuckGo's sustained growth to a 0.86% global search market share by mid-2025, attracting users prioritizing non-tracking models without disrupting broader industry viability. Right-leaning viewpoints counter surveillance capitalism narratives by advocating property rights frameworks for , positing that treating data as individually owned assets—exchangeable via —promotes and counters regulatory overreach that could entrench incumbents under guise of protection. This approach highlights how profit-driven tracking incentivizes value creation, such as personalized services, over coercion, with competitive ad markets featuring diverse intermediaries preventing any singular behavioral monopoly. While acknowledging commercialization risks, these perspectives prioritize empirical market outcomes, where user choices and exit options mitigate power imbalances more effectively than alarmist recharacterizations.

Technical Vulnerabilities

Probabilistic cross-device tracking, relying on correlations of shared attributes such as IP addresses, timestamps, and behavioral signals, yields accuracy estimates up to 97.3% but incurs misattribution errors from factors like dynamic IP assignments or household-shared networks, with discrepancies often falling in the 2-5% range based on controlled measurements. These graph mismatches disrupt event chaining, where unrelated devices are erroneously linked, inflating false positives in identity resolution by conflating proximity-based signals with true user continuity. Device fingerprinting techniques, aggregating browser and hardware traits for cross-device linkage, falter against evasion methods including user-agent randomization or proxy alterations, which fragment fingerprints and yield linkage failures rates exceeding 10% in adversarial scenarios. Such evasions stem from the method's dependence on static or semi-static attributes vulnerable to deliberate perturbation, analogous to noise in that erodes probabilistic model confidence. Ultrasonic audio s, utilized for proximity-based cross-device signaling in systems like SilverPush's SDK, expose flaws in unverified handling and unencrypted transmission, permitting replay attacks where intercepted signals (18-24 kHz range) are rebroadcast to fabricate false device associations. on nearby devices can passively capture these inaudible emissions, enabling side-channel exfiltration of tracking IDs without checks, as demonstrated in analyses of injection vectors. These ultrasonic exploits, while enabling unintended correlations across unlinked hardware, proved rare in deployment; SilverPush phased out the feature by 2016 through code excision and signal suppression updates, reducing incidence to negligible levels in subsequent audits. Mitigation via hardened protocols, such as encrypted beacons or frequency hopping, parallels defenses in RF-based systems, underscoring that cross-device vulnerabilities reflect standard engineering trade-offs in distributed data linkage rather than unique systemic frailties.

Key Global Regulations

The European Union's (GDPR), effective May 25, 2018, classifies cross-device tracking as of requiring a lawful basis such as explicit consent under Article 6, particularly when involving profiling or across devices that infer user identity. Enforcement by national data protection authorities has imposed fines for inadequate consent in tracking practices, with total penalties reaching approximately $6.4 billion by January 2025, compelling companies to enhance transparency and limit deterministic linking reliant on identifiers like emails or logins. However, studies post-GDPR reveal an initial short-term decline in tracker deployment followed by a rebound, as probabilistic techniques—using signals like device fingerprints or behavioral patterns—persist due to challenges in proving they constitute without explicit linkage. In the United States, the , effective January 1, 2020, and expanded by the from January 1, 2023, empowers consumers to opt out of the "sale" or "sharing" of , explicitly targeting cross-context behavioral advertising that spans devices and platforms. The CPRA closes prior gaps by regulating sharing for advertising purposes as distinct from internal analytics, requiring businesses to honor global opt-out signals and imposing penalties up to $7,500 per intentional violation, which has disrupted data brokers' deterministic cross-device graphs and accelerated adoption of consented, first-party data models. Beyond these, Brazil's Lei Geral de Proteção de Dados (LGPD), enforced since August 2020, mandates for non-essential data processing including tracking, with fines up to 2% of Brazilian revenue, though enforcement remains nascent and has minimally altered global deployment patterns. China's Personal Information Protection Law (PIPL), effective November 2021, requires separate for sensitive automated processing and restricts cross-border data flows, impacting multinational trackers but primarily through localized compliance rather than broad curtailment of probabilistic methods. These regulations collectively enforce and mechanisms, reducing overt deterministic cross-device practices in compliant jurisdictions while spurring shifts to less detectable alternatives, underscoring enforcement's uneven effect on innovation versus persistence.

Platform-Specific Policies

Apple introduced App Tracking Transparency (ATT) with the iOS 14.5 update on April 26, 2021, requiring apps to obtain explicit user consent before accessing the (IDFA) for cross-app and cross-site tracking. To support ad attribution without relying on personalized identifiers, Apple developed SKAdNetwork, a framework that aggregates conversion data in a privacy-preserving manner, limiting granular reporting to coarse postbacks with randomized timing and no user-level identifiers. These measures prioritize device-level privacy by restricting cross-device linkage unless users opt in, though critics argue ATT enables by consolidating control over app distribution and data flows through Apple's dominance, as evidenced by a €150 million fine from France's competition authority in April 2025 for abusing its position in implementing privacy features that disadvantaged alternative tracking providers. Google's Consent Mode v2, rolled out in late November 2023, extends consent signaling to include ad and ad user parameters alongside existing storage consents, enabling adjusted modeling of user across devices and sessions in regions with strict consent requirements like the EEA, while becoming mandatory for certain ad products by March 2024. This integrates with efforts to phase out third-party cookies in Chrome, originally targeted for early 2025 but ultimately extended indefinitely amid unresolved competition concerns and the discontinuation of the initiative in October 2025, preserving some cross-device utility through first-party and server-side signals rather than fully eliminating tracking capabilities. Meta implemented Aggregated Event Measurement following Apple's updates, allowing limited reporting of up to eight prioritized events per domain from + devices via privacy-safe aggregation that deduplicates signals and withholds raw user data, even from opted-out users, to facilitate cross-app and web-to-app attribution without direct identifiers. For cross-device scenarios, Meta's policies incorporate opt-in mechanisms for reporting ad interactions across devices leading to conversions, as outlined in their cross-device reporting tools, with enhanced controls introduced in 2024 to align with regulatory demands for explicit user approval on beyond single-device boundaries. These approaches balance measurement needs by relying on modeled aggregates and domain-verified events, though they reduce precision in cross-device graphs compared to pre- identifier-based methods.

Enforcement and Compliance Challenges

Enforcing regulations on cross-device tracking faces significant hurdles due to the inherent opacity of probabilistic methods, which infer user identities across devices using statistical patterns from behavioral data rather than explicit identifiers like . These techniques, often invisible to consumers and regulators alike, complicate detection and verification of compliance, as they do not rely on user-controllable signals and can evade traditional mechanisms. In practice, this probabilistic nature allows circumvention of rules mandating transparency and options, as linkages are derived indirectly from aggregated signals, making it difficult for authorities to or prove violations without deep technical access. Recent enforcement actions underscore persistent lags in oversight. In 2024, the U.S. (FTC) initiated probes and settlements against data brokers such as Mobilewalla and Gravy Analytics for unlawfully selling sensitive location data that enables cross-device linkage, prohibiting future sales and imposing data deletion requirements. These cases highlight reactive rather than preventive efficacy, as similar practices continued despite prior warnings, with the FTC noting inadequate safeguards against sensitive inferences like visits to medical facilities. Globally, under the EU's GDPR, compliance has imposed measurable costs, equivalent to a 20% average increase in expenses for firms, driven by requirements for consent audits and data minimization. Yet, unauthorized data trading persists in underground markets, where personal identifiers are commodified despite regulatory bans, exacerbating enforcement gaps through jurisdictional mismatches and anonymous exchanges. Self-regulatory initiatives partially address these voids. The Network Advertising Initiative (NAI) enforces a voluntary for members, mandating , mechanisms, and restrictions on sensitive data use in interest-based , including cross-device scenarios. This framework supplements statutory rules by promoting standardized opt-outs and accountability, though adherence relies on industry participation rather than mandates. While such regulations demonstrably curb egregious practices like unconsented , they risk overreach by imposing disproportionate burdens on smaller advertising entities, which allocate higher per-employee resources to compliance than larger competitors, potentially stifling and market entry.

Future Directions

Emerging Technologies

In response to evolving privacy regulations such as the phase-out of third-party , developers are advancing algorithms for probabilistic identity resolution in cross-device tracking, enabling linkages across devices with mathematical privacy guarantees like to minimize re-identification risks. These AI-driven models analyze behavioral signals, such as browsing patterns and timing correlations, to construct user identity graphs that achieve reliable matching without relying on persistent identifiers. For instance, post-cookie attribution systems incorporate ML to infer cross-device journeys, supporting advertisers in maintaining targeting efficacy while navigating consent requirements. Google's policy shift effective February 16, 2025, permits the use of device fingerprinting techniques within Chrome for advertising purposes, marking a reversal from prior restrictions and facilitating enhanced cross-device measurement by aggregating signals like browser configurations, IP addresses, and hardware attributes. This evolution underscores the persistence of fingerprinting as a core method, allowing platforms to link user activities across ecosystems despite regulatory scrutiny from bodies like the UK's , which has cautioned against unchecked deployment. Critics note that such techniques enable non-consensual persistent tracking, potentially spanning smart devices and browsers. Decentralized identity solutions leveraging , particularly (SSI) frameworks, are undergoing enterprise pilots to provide verifiable cross-device credentials without centralized intermediaries. Financial institutions tested SSI in to reduce verification costs by up to USD 35 billion annually through blockchain-anchored digital wallets that users control across devices. These systems employ for selective disclosure, potentially adapting tracking by enabling consented, tamper-proof linkages in sectors like and , though remains a challenge in widespread adoption.

Privacy-Enhancing Alternatives

enables model training across devices without centralizing raw user data, as updates are computed locally and aggregated server-side, thereby reducing exposure to cross-device linkage risks. Apple's implementation, combining with for tasks like , ensures noise addition to updates, limiting attacks while maintaining , as demonstrated in their 2023 framework for end-to-end models. This approach minimizes data transmission but requires sufficient device participation to achieve convergence, with empirical evaluations showing privacy gains through data minimization principles. Virtual private networks (VPNs) and the Tor network obscure IP addresses and routing paths, impeding trackers reliant on network signals for cross-device correlation. VPNs encrypt traffic and mask origins, effectively blocking IP-based fingerprinting in many scenarios, while Tor's multi-hop onion routing adds layers of indirection, enhancing resistance to surveillance. Studies indicate these tools can reduce observable tracking signals by altering identifiers, though they falter against non-IP methods like browser fingerprints or device graphs. Browser extensions such as employ filter lists to block third-party and scripts, targeting common tracking mechanisms with high efficacy against known vectors. Independent tests reveal neutralizes a substantial portion of cookie-based trackers, often exceeding 80% blockage rates for ad-related domains, though server-side and fingerprinting techniques evade full mitigation. Privacy tools broadly, including ad blockers and search engines like , have been shown to halve persistent tracking from dominant providers in controlled audits. These alternatives introduce trade-offs, notably diminished accuracy, as aggregated or obscured data yields less granular user profiles for services like recommendations. Federated methods preserve some utility via local but aggregate noise can degrade model performance by 10-20% in privacy-sensitive domains without compensatory layers. Network obfuscators and blockers similarly disrupt service tailoring, prompting the -privacy where enhanced correlates with reduced in targeted content. Critically, such technologies often consolidate control among platform incumbents, as proprietary implementations like Apple's retain lock-in without obviating underlying data dependencies for global model refinement.

Potential Industry Evolutions

The advertising industry reliant on cross-device tracking is evolving toward greater emphasis on first-party data collected directly by publishers and retailers, diminishing dependence on third-party data brokers amid the phase-out of and heightened scrutiny. This shift incentivizes platforms to build proprietary identity graphs linking user behaviors across devices via logged-in sessions and purchase histories, enhancing targeting precision without intermediary vulnerabilities. Retail media networks exemplify this trend, with U.S. revenues projected to grow 20% in 2025, outpacing total ad market expansion and capturing a larger share of digital spend through owned first-party signals. Such networks, operated by entities like , reported 27% year-over-year ad revenue increases in fiscal 2024, driven by on-site and off-site extensions that leverage transactional data for cross-device attribution. Regulatory pressures, including the European Union's €2.95 billion fine against in September 2025 for anti-competitive ad tech practices, may accelerate consolidation of tracking capabilities within dominant ecosystems like Apple's and Google's Android, where privacy gates such as App Tracking Transparency limit third-party access but preserve intra-platform graphing. These rules, while aiming to curb monopolies, risk entrenching "walled gardens" by raising compliance costs for smaller players, potentially reducing broker diversity as economic scale favors incumbents with vast user bases for probabilistic and deterministic matching. However, agnostic or open identity frameworks—linking emails, device IDs, and logins across vendors—could promote by enabling interoperable graphs, as seen in solutions from platforms like that unify cross-device views without proprietary lock-in. Cross-device tracking remains integral to digital economies, with global ad revenues forecasted to exceed $1 trillion in 2025, underscoring economic imperatives that render outright bans improbable despite advocacy for restrictions. First-principles analysis of incentives reveals that advertisers' need for scalable, ROI-driven —tied to trillions in cumulative value—will sustain innovations in consented ecosystems over disruptive prohibitions, as evidenced by persistent growth in commerce media at a 21% CAGR through 2027. This realism tempers expectations of radical overhaul, favoring adaptive structures that balance efficacy with regulatory realities rather than idealistic curtailments.

References

  1. https://www.adjust.com/glossary/cross-device-tracking/
  2. https://www.thebridgecorp.com/deterministic-vs-probabilistic/
  3. https://digiday.com/media/deterministic-vs-probabilistic-cross-device-tracking-explained-normals/
  4. https://amplitude.com/blog/cross-device-tracking
  5. https://www.ftc.gov/system/files/documents/reports/cross-device-tracking-federal-trade-commission-staff-report-january-2017/ftc_cross-device_tracking_report_1-23-17.pdf
  6. http://www.cs.columbia.edu/~jebara/papers/Cross_Device_Tracking_2017.pdf
  7. https://www.internetandtechnologylaw.com/ccpa-gdpr-cross-device-tracking/
  8. https://www.researchgate.net/publication/330037184_Cross-Device_Tracking_Systematic_Method_to_Detect_and_Measure_CDT
  9. https://usercentrics.com/guides/marketing-measurement/cross-device-tracking/
  10. https://stape.io/blog/cross-device-tracking
  11. https://porchgroupmedia.com/blog/25-amazing-omnichannel-statistics-every-marketer-should-know/
  12. https://www.lotame.com/resources/benefits-cross-device-marketing/
  13. https://www.heap.io/topics/cross-device-tracking/
  14. https://impact.com/affiliate/cross-device-tracking/
  15. https://www.mparticle.com/blog/probabilistic-vs-deterministic/
  16. https://www.linkedin.com/pulse/beginners-guide-deterministic-probabilistic-tracking-mporium-stephen
  17. https://help.funnel.io/en/articles/11985410-cross-device-tracking-deterministic-and-probabilistic-approaches
  18. https://www.heap.io/topics/cross-device-tracking
  19. https://www.go.resul.io/blog/deterministic-vs-probabilistic-matching-accuracy-or-scale
  20. https://www.northbeam.io/blog/probabilistic-vs-deterministic-attribution-marketing
  21. https://www.herm.io/blog/the-truth-about-cross-device-identity-resolution-in-2025
  22. https://audiencelab.io/resources/workshop/identity-resolution-understanding-deterministic-vs-probabilistic-matching/
  23. https://www.researchgate.net/publication/393805577_Constructing_Cross-Device_Ad_Attribution_Models_for_Integrated_Performance_Measurement
  24. https://www.cookieyes.com/blog/internet-cookies/
  25. https://www.danny.engineering/article/magic-cookies-a-brief-history-of-browser-cookies
  26. https://www.researchgate.net/publication/200110525_Detecting_Session_Boundaries_from_Web_User_Logs
  27. https://timelines.issarice.com/wiki/Timeline_of_online_advertising
  28. https://cdn.aaai.org/ocs/1070/1070-6009-1-PB.pdf
  29. https://www.zscaler.com/blogs/security-research/demystifyingabusing-flash-cookies
  30. https://transition.fcc.gov/national-broadband-plan/broadband-adoption-in-america-paper.pdf
  31. https://www.usenix.org/system/files/conference/usenixsecurity16/sec16_paper_lerner.pdf
  32. https://www.smartinsights.com/mobile-marketing/mobile-marketing-analytics/mobile-usage-statistics-2010-2015/
  33. https://addict-mobile.com/en/user-acquisition-attribution/
  34. https://help.adjust.com/en/article/attribution-methods
  35. https://www.adexchanger.com/data-exchanges/how-much-cross-device-clout-do-facebook-and-google-actually-have/
  36. https://www.ftc.gov/news-events/events/2015/11/cross-device-tracking
  37. https://www.ftc.gov/news-events/news/press-releases/2015/03/ftc-host-workshop-cross-device-tracking-nov-16
  38. https://www.ftc.gov/system/files/documents/reports/data-brokers-call-transparency-accountability-report-federal-trade-commission-may-2014/140527databrokerreport.pdf
  39. https://en-gb.facebook.com/business/news/insights/mobile-advertising-now-drives-over-half-of-all-campaign-conversions
  40. https://www.facebook.com/business/news/cross-device-measurement
  41. https://weareaxm.com/ideas/the-impact-of-ios14-5-and-app-tracking-transparency-att/
  42. https://applemagazine.com/app-tracking-transparency-cuts-u-s-tracking-rate/
  43. https://www.linkedin.com/posts/danielmcc_really-enjoyed-this-interview-with-mike-linton-activity-7371164152658149376-dJI1
  44. https://www.rhsmith.umd.edu/research/small-businesses-take-big-hit-apples-privacy-regulation
  45. https://privacysandbox.google.com/blog/cookie-countdown-2023oct
  46. https://www.cookieyes.com/blog/third-party-cookies-going-away/
  47. https://seers.ai/blogs/google-cookie-deprecation/
  48. https://support.google.com/analytics/answer/10089681?hl=en
  49. https://www.owox.com/blog/articles/key-benefits-of-google-analytics-4
  50. https://seizemarketingagency.com/how-to-leverage-google-analytics-4/?srsltid=AfmBOorNOd91Z5NVjVNNaXCafzvbgXNc5A9gSosR6G2LiejSKYhffn9v
  51. https://www.shopify.com/enterprise/blog/cross-device-ad-targeting
  52. https://www.mile.tech/blog/deterministic-and-probabilistic-ids
  53. https://www.opensend.com/post/cross-device-tracking
  54. https://xenoss.io/blog/cookieless-solutions
  55. https://centro.net/wp-content/uploads/2018/09/CI_WS_Cross-Device.pdf
  56. https://canopusinfosystems.com/cross-device-tracking-and-attribution/
  57. https://iabtechlab.com/wp-content/uploads/2024/05/Identity-Solutions-Guidance-FINAL.pdf
  58. https://petsymposium.org/popets/2017/popets-2017-0020.pdf
  59. https://www.oracle.com/za/a/ocom/docs/cx-activation-vertical-playbook-2020.pdf
  60. https://imai.fas.harvard.edu/research/files/linkage.pdf
  61. https://datadome.co/learning-center/browser-fingerprinting-techniques/
  62. https://seon.io/resources/browser-fingerprinting/
  63. https://web.dev/learn/privacy/fingerprinting
  64. https://arxiv.org/pdf/1905.05337
  65. https://www.wired.com/2016/11/block-ultrasonic-signals-didnt-know-tracking/
  66. https://patents.google.com/patent/US9024998B2/en
  67. https://patents.google.com/patent/US20150215668A1/en
  68. https://dl.acm.org/doi/10.1145/3176258.3176313
  69. https://www.comparitech.com/blog/information-security/block-ultrasonic-tracking-apps/
  70. https://phys.org/news/2018-05-ultrasound-firewall-mobile.html
  71. https://www.blackhat.com/eu-16/briefings/schedule/#talking-behind-your-back-attacks-and-countermeasures-of-ultrasonic-cross-device-tracking-4864
  72. https://reyammer.io/publications/2017_pets_ultrasound.pdf
  73. https://doubleclick-advertisers.googleblog.com/2015/06/cross-device-conversion-metrics-come-to.html
  74. https://www.optimizesmart.com/cross-device-attribution-reports-google-adwords/
  75. https://cypressnorth.com/paid-search-marketing/google-marketing-live-2025-the-biggest-changes-coming-to-google-ads-how-theyll-impact-you/
  76. https://ppc.land/google-unveils-journey-aware-bidding-to-optimize-full-customer-paths/
  77. https://netconsult.ae/google-ads-in-2025/
  78. https://haus.io/blog/the-meta-report-lessons-from-640-haus-incrementality-experiments
  79. https://www.trackbee.io/blog/metas-incremental-attribution-a-step-forward-but-not-the-complete-picture
  80. https://support.google.com/analytics/answer/9213390?hl=en
  81. https://mercadoglobalmedia.com/ga4-user-id-tracking/
  82. https://swetrix.com/blog/cross-device-tracking
  83. https://stratoflow.com/how-netflix-recommendation-algorithm-work/
  84. https://www.trypropel.ai/resources/customer-retention-strategies-proven-netflix-techniques-to-reduce-churn
  85. https://www.amazon.com/gp/help/customer/display.html?nodeId=G5LD93RTW3MMDKCR
  86. https://www.ruleranalytics.com/blog/analytics/google-analytics-user-tracking/
  87. https://www.sciencedirect.com/science/article/pii/S2215016124001377
  88. https://www.tigergraph.com/glossary/fraud-detection-with-graph/
  89. https://www.paypal.com/us/enterprise/fraud-protection-advanced
  90. https://aimerce.ai/topics/cross-device-tracking-mastery-guide-47
  91. https://blog.quuu.co/cross-device-tracking-challenges-explained/
  92. https://www.oberlo.com/statistics/digital-ad-spend
  93. https://fpf.org/wp-content/uploads/2015/11/FPF_FTC_CrossDevice_F_20pg-3.pdf
  94. https://www.gspann.com/resources/blogs/cross-channel-and-cross-device-tracking-creates-better-customer-experiences-and-saves-millions-in-ad-spends/
  95. https://support.apple.com/guide/mac-help/intro-to-continuity-mchl1d734309/mac
  96. https://www.apple.com/macos/continuity/
  97. https://journalwjarr.com/sites/default/files/fulltext_pdf/WJARR-2025-1079.pdf
  98. https://link.springer.com/article/10.1007/s11002-024-09726-6
  99. https://www.darden.virginia.edu/sites/default/files/inline-files/Raveendhran_Fast_2024.pdf
  100. https://www.precedenceresearch.com/digital-ad-spending-market
  101. https://www.iopex.com/blog/how-ai-transforming-the-adtech-industry
  102. https://builtin.com/articles/top-data-broker-companies
  103. https://finance.yahoo.com/news/data-broker-industry-analysis-report-090900272.html
  104. https://redstagfulfillment.com/2010s-ecommerce-growth-decade/
  105. https://www.go-globe.com/from-brick-to-click-the-evolutionary-journey-of-ecommerce/
  106. https://www.wilmerhale.com/en/insights/blogs/wilmerhale-privacy-and-cybersecurity-law/ftc-staff-issues-long-awaited-cross-device-tracking-report
  107. https://oag.ca.gov/privacy/ccpa
  108. https://www.cooley.com/news/insight/2025/2025-05-30-cppa-enforcement-heats-up-recent-enforcement-actions-and-their-implications
  109. https://www.thomsonreuters.com/en-us/posts/government/identity-theft-drivers/
  110. https://seon.io/resources/synthetic-identity-fraud-prevention-and-detection/
  111. https://dl.acm.org/doi/10.1145/3706598.3714318
  112. https://pmc.ncbi.nlm.nih.gov/articles/PMC7349636/
  113. https://objectfirst.com/guides/data-security/types-of-data-breaches/
  114. https://www.bea.gov/research/papers/2018/value-data-theres-no-such-thing-free-lunch-digital-economy
  115. https://www.bea.gov/sites/default/files/papers/20190220ValueofDataLiNireiYamanaforBEAworkingpaper.pdf
  116. https://www.witsa.org/news/big-techs-free-online-services-arent-costing-consumers-their-privacy
  117. https://thebaffler.com/latest/capitalisms-new-clothes-morozov
  118. https://link.springer.com/article/10.1007/s13347-024-00804-1
  119. https://backlinko.com/ad-blockers-users
  120. https://seranking.com/blog/top-search-engines/
  121. https://www.aei.org/research-products/report/the-semantics-of-surveillance-capitalism-much-ado-about-something/
  122. https://laweconcenter.org/resources/the-antitrust-assault-on-ad-tech-a-law-economics-critique/
  123. https://www.adexchanger.com/data-driven-thinking/when-evaluating-cross-device-graph-technology-look-beyond-match-accuracy/
  124. https://arxiv.org/html/2502.01608v1
  125. https://www.nccgroup.com/media/00rjhgp1/_ultrasound-whitepaper-v10.pdf
  126. https://mlsec.org/docs/2017a-eurosp.pdf
  127. https://www.reform.app/blog/cross-border-data-privacy-key-enforcement-issues
  128. https://www.sciencedirect.com/science/article/pii/S0167811625000229
  129. https://policyreview.info/articles/analysis/and-after-gdpr-tracking-mobile-apps
  130. https://calawyers.org/business-law/digital-trackers-data-protection-how-the-cpra-closes-ccpa-gaps-in-addressing-cross-contextual-behavioral-tracking/
  131. https://www.adexchanger.com/data-driven-thinking/what-does-the-cpra-mean-for-behavioral-advertising/
  132. https://piwik.pro/privacy-laws-around-globe/
  133. https://securiti.ai/privacy-laws/
  134. https://www.adjust.com/glossary/app-tracking-transparency/
  135. https://techcrunch.com/2021/04/26/apples-app-tracking-transparency-feature-has-arrived-heres-what-you-need-to-know/
  136. https://www.appsflyer.com/glossary/skadnetwork/
  137. https://www.adjust.com/glossary/skadnetwork-skan/
  138. https://www.clearyantitrustwatch.com/2025/04/french-competition-authority-fines-apple-e150-million-for-abusive-implementation-of-privacy-framework/
  139. https://digiday.com/marketing/after-the-e150-million-fine-apples-att-faces-its-hardest-questions-yet/
  140. https://www.uniconsent.com/blog/google-consent-mode-v2-explained
  141. https://www.cookieyes.com/knowledge-base/google-consent-mode/what-is-google-consent-mode-v2/
  142. https://support.google.com/google-ads/answer/14762010?hl=en
  143. https://searchengineland.com/google-officially-shuts-down-privacy-sandbox-463561
  144. https://adtelligent.com/blog/ad-tech-insights/how-third-party-cookies-elimination-will-affect-programmatic-ecosystem/
  145. https://www.facebook.com/business/help/721422165168355
  146. https://www.facebook.com/business/help/1356268495231843
  147. https://www.facebook.com/business/help/266118906912295
  148. https://secureprivacy.ai/blog/meta-consent-mode-explained-2025
  149. https://www.ftc.gov/news-events/news/press-releases/2024/12/ftc-takes-action-against-mobilewalla-collecting-selling-sensitive-location-data
  150. https://www.ftc.gov/news-events/news/press-releases/2024/12/ftc-takes-action-against-gravy-analytics-venntel-unlawfully-selling-location-data-tracking-consumers
  151. https://mitsloan.mit.edu/ideas-made-to-matter/gdpr-reduced-firms-data-and-computation-use
  152. https://www.researchgate.net/publication/370029546_The_black_market_of_personal_data_financial_legal_and_social_aspects
  153. https://thenai.org/self-regulatory-framework/
  154. https://thenai.org/wp-content/uploads/2021/07/NAI_Code15encr.pdf
  155. https://www.uschamber.com/small-business/a-majority-of-small-businesses-say-regulations-are-hindering-growth
  156. https://theconversation.com/a-bipartisan-data-privacy-law-could-backfire-on-small-businesses-2-marketing-professors-explain-why-234771
  157. https://madgicx.com/blog/machine-learning-for-multi-touch-attribution
  158. https://www.emarketer.com/content/google-pivot-digital-fingerprinting-enable-better-cross-device-measurement
  159. https://ico.org.uk/about-the-ico/media-centre/news-and-blogs/2024/12/our-response-to-google-s-policy-change-on-fingerprinting/
  160. https://www.forbes.com/sites/zakdoffman/2024/12/21/forget-chrome-google-will-start-tracking-you-and-all-your-smart-devices-in-8-weeks/
  161. https://www.mordorintelligence.com/industry-reports/blockchain-identity-management-market
  162. https://www.bobsguide.com/self-sovereign-identity-ssi-on-blockchain-reshaping-trust-and-compliance/
  163. https://www.ibm.com/think/topics/federated-learning
  164. https://machinelearning.apple.com/research/fed-learning-diff-privacy
  165. https://docs.cloud.google.com/architecture/cross-silo-cross-device-federated-learning-google-cloud
  166. https://www.expressvpn.com/blog/prevent-cross-site-tracking/
  167. https://www.privateinternetaccess.com/blog/tor-vs-vpn/
  168. https://www.pcworld.com/article/2791638/why-you-should-not-use-a-vpn-together-with-the-tor-network.html
  169. https://cybernews.com/best-ad-blockers/ublock-origin-review/
  170. https://negg.blog/en/ublock-origin-advanced-ad-blocking/
  171. https://hackread.com/new-study-google-tracking-persists-privacy-tools/
  172. https://proceedings.mlr.press/v162/bietti22a/bietti22a.pdf
  173. https://www.sciencedirect.com/science/article/abs/pii/S1084804520302794
  174. https://cmr.berkeley.edu/2025/02/balancing-personalized-marketing-and-data-privacy-in-the-era-of-ai/
  175. https://machinelearning.apple.com/research/scenes-differential-privacy
  176. https://www.emarketer.com/content/how-brands-adapting-post-cookie-world-with-first-party-data
  177. https://hbr.org/sponsored/2022/09/with-first-party-data-marketers-are-finally-in-the-drivers-seat
  178. https://www.nielsen.com/insights/2025/future-retail-media/
  179. https://www.marketingbrew.com/stories/2025/03/17/how-the-largest-retail-media-networks-are-stacking-up
  180. https://www.reuters.com/legal/litigation/google-hit-with-345-billion-eu-antitrust-fine-over-adtech-practices-2025-09-05/
  181. https://www.ainvest.com/news/eu-eur3-billion-google-adtech-fine-catalyst-market-consolidation-digital-advertising-2509/
  182. https://aerospike.com/blog/what-is-identity-graph/
  183. https://www.thetradedesk.com/our-demand-side-platform/identity-solutions
  184. https://www.adexchanger.com/content-studio/agnostic-identity-graphs-are-the-future-of-identity-resolution/
  185. https://www.wppmedia.com/news/tyny-midyear-2025
  186. https://www.emarketer.com/content/worldwide-ad-spending-forecast-2025
  187. https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights/the-evolution-of-commerce-media-navigating-a-new-era-in-advertising
Add your contribution
Related Hubs
User Avatar
No comments yet.