The Impact of Digital Acts on Audience Targeting Strategies

Today there is a need to adopt new approaches to audience creation, which deviate significantly from the strategies employed in past years. This need is rooted in regulatory changes, most notably the introduction of the Digital Markets Act, a regulation launched by the European Commission, which came into force on 6 March 2024.

The DMA is not directly linked to audience formation methodologies, but it does affect them and in the following paragraphs we will see how.

The DMA aims on the one hand to protect users’ privacy and on the other hand to ensure fairer and more equal competition in digital markets.

One of the most relevant novelties introduced by the DMA concerns the figure of gatekeepers: guardians who regulate access to digital functionalities, who must meet specific requirements to be classified as such. Criteria include: being of a significant size, with revenues exceeding EUR 7 billion in the last three years; having extensive access to user data; and having a solid position in the market.

The companies currently identified as gatekeepers are: Alphabet (Google’s holding company), Amazon, Apple, ByteDance (the holding company that controls TikTok), Meta and Microsoft. These six giants manage a total of 22 ‘Core Platform Services’, which cover different sectors: marketplaces, social networks, online advertising services, operating systems, search engines and browsers.

For the DMA, the importance of gatekeepers lies in their ability to significantly influence market dynamics, as their control over certain platforms can restrict access to smaller competitors. To counteract this trend and promote a fairer market environment, gatekeepers must fulfil certain obligations, namely:

• Ensure greater openness on their platforms. For instance, if a user searches for a specific location via Google on Chrome, he/she must be able to receive results that include not only Google Maps, but also similar services offered by other companies, so as to reduce barriers to entry for potential competitors;
• Obtaining users’ explicit consent before using their data. Responsibility for the collection and processing of data is no longer the sole prerogative of companies, gatekeepers must ensure that third-party platforms also operate in compliance with the consent provided by users.

Google, with regard to compliance with the DMA, has indicated that from 2024 it will be essential to obtain explicit consent before sending data to platforms. To this end, it has started to improve and update consent management technologies that, although developed before 2024, have become crucial for compliance with the new regulations: this is the case with consent mode.

Consent Mode: what it is and how it works

Consent Mode is a technology introduced in September 2020, the adoption of which was not mandatory but strongly recommended. However, with the arrival of the Digital Markets Act, it has taken on new connotations and requirements.

Consent Mode enables the sending of cookieless signals to Google, i.e. without using cookie-based technologies, for users who have not given their consent. This will result in both observed and modelled conversions. In the second case, Google models consent-free and anonymised data through the application of machine learning techniques, so as to estimate the number of users who actually completed a conversion.

The key parameters that support the Consent Mode structure are ‘ad_storage’, which handles the storage of information related to advertising, and ‘analytics_storage’, which handles information related to data analysis

These parameters indicate whether or not the user has consented to be tracked for marketing purposes. If the user grants permission, the value will be ‘granted’; if not, it will be ‘denied’.

In 2024, with the introduction of the DMA, two further advertising-related parameters were introduced:

• ”ad_user_data’ which governs consent for user data in advertising;
• ‘ad_personalisation’ which governs consent for personalised advertising.

These two parameters cannot be separated from ‘ad_storage’, as they are considered a subset of it. Their implementation has significant implications for tag management and audience creation and delivery to platforms. When a user grants consent, tags function normally, activating first-party tracking cookies, which are unaffected by the cookieless context. This enables effective tracking: the conversions obtained are the real ones observed by the advertising measurement platforms.On the other hand, when consent is denied, tags are activated that do not store cookies and only send signals indicating the denied consent status. In the case of conversion, a signal is sent, collecting anonymous information, e.g. on the technology used by the user, without further granular details. As far as Analytics is concerned, signals are sent concerning visited pages and recorded events, also without cookies. This data is then modelled by Google to provide platforms with an estimate that, while not completely accurate, comes close to the platforms’ actual conversion rates.Not implementing Consent Mode will result in:

• The gradual deactivation of remarketing audiences, as Google will stop transmitting data on users who have not activated Consent Mode;
• The difficulty in effectively building marketing strategies based on lookalike audiences, which are essential for Demand Generation campaigns;
• The inability to execute dynamic remarketing campaigns, due to the lack of data and signals;
• The inaccuracy in the calculation of conversions, as it is not possible to model the conversions of all users who have not provided consent.

There are two ways of implementing Consent Mode:

• Basic implementation, tags are not activated unless the user explicitly provides consent for cookies. If the user visits a site and gives consent, tags will not be deployed until it is clear whether they wish to be tracked. Once consent has been obtained, tracking starts with additional information related to Consent Mode, allowing the data of users who have not given consent to be modelled or the observed data of those who have.
• Advanced Implementation, all Google tags are activated regardless of user consent, as Consent Mode modulates the pings and information sent. An advantage of the advanced approach is that modelling is not only limited to conversions in Google Ads, but also extends to user behaviour in Google Analytics, including modelling the traffic data of users who have not provided consent. Such machine learning-based modelling work requires a considerable volume of data, at least 700 ad clicks over 7 continuous days per domain or country, in order for Google to effectively model the data and provide useful information for a correct perception of conversions, sessions and traffic.

Solutions to the challenges generated by the Digital Acts

Against this backdrop, it has become crucial for companies to identify and implement effective strategies that not only meet current regulations, but also allow them to capitalise in this new, more open and competitive market scenario.

Dynamic Consensus with the Modern Data Stack

The process of managing user consent has been driven by stringent regulations such as the GDPR, which introduced mandatory cookie banners. These regulations have often generated extensive discussion and reflection among Data Protection Officers (DPOs) and companies’ legal departments, leading to the need to balance legal compliance with business impact. In some cases, this has involved cost-benefit assessments, leading some companies to run the risk of not fully adhering to cookie and consent management regulations.

Currently, ignoring consent collection is tantamount to foregoing digital advertising. It is therefore crucial to be compliant and understand how to handle consents expressed in different ways in different contexts. For example, in the case of users who express consent to marketing and measurement via a loyalty card, but refuse consent on the company’s website.

This is the challenge for modern companies: to be dynamic and resilient, especially those operating in both the physical and digital worlds, with different points of user interaction.

One solution is to centralise data management in a single Cloud Data Warehouse. This approach allows transactional and behavioural data to be collected and integrated with consent data, creating a platform that is always up-to-date based on the user’s last consent and able to react appropriately.

Having a single, centralised point for consent management places the company in an advantageous position with respect to the obligations imposed by the Digital Acts.

Focus on First Party Audiences

The Digital Acts impacted on the process of linking services offered by large digital players, making it impossible to link user behaviour between different properties without explicit user consent.

Un-linking does not necessarily lead to a degradation of audiences, but it is widely recognised among analysts and digital platforms themselves that the quality of internal audiences, such as those that could be gathered within YouTube or so-called ‘Life Events’ audiences, will decline. This phenomenon is aggravated by the phasing out of third-party cookies

To maintain quality audiences, it is essential to obtain user consent and create first-party audiences. Searching online for best practices for digital acts and how to react to unlinking, one can find first-party audiences among the first strategies. Google, for instance, has organised numerous webinars on this topic.

To effectively build first-party audiences, in addition to consensus, it is crucial to have a robust data aggregation point. In addition, AI algorithms play a key role in this process, allowing the customer base to be dynamically enriched and segmented. An example is the RFM algorithm to identify the most important customers. Or specific algorithms to analyse user interests. This data, once enriched and modelled, can then be used to build ‘seed audiences’ that serve as the basis for lookalike campaigns.

In the past there has been some confusion about lookalike campaigns due to Google’s decision to deprecate ‘similar audiences’. It actually replaced them with a new set of algorithms that exploit similarity in different ways.

Bytek Prediction Platform uses artificial intelligence and first-party data to cluster users, creating well-defined audiences that can be continuously sent to advertising platforms for optimised use in campaigns.

Customer Match

Customer matching is a methodology that allows companies to use first-party data, such as a hashed email address, to identify users, send them to advertising platforms such as Google, Meta or TikTok, which then check whether the hashed email address is present in their systems, without revealing the specific identity of the individual, in order to preserve privacy. If the email is recognised, they allow companies to run retargeting campaigns.

This technique has been seen as a pioneering strategy for digital marketing, as it avoids the use of third-party cookies, benefiting from a direct 1:1 relationship between platforms and users. However, as we have seen, the introduction of DMA has introduced new challenges: Google and other platforms require explicit consent to show personalised advertising via customer match.

In order to implement customer match, if one decides to transfer data to Google via a CSV file, the platform requires that advertisers confirm that they have specific consent to personalise ads. This requires a double confirmation via a banner. The responsibility for ensuring compliance therefore lies with the advertisers. If, on the other hand, data is transmitted via APIs, the consent for ad customisation must be specified for each user, indicating whether and to what types of data the user has given consent.

It is therefore necessary to rethink how consent itself should be handled. By using tools like BigQuery to centralise data, audiences can be dynamically updated according to changes in user consent. In the absence of this specific consent information in Google, the corresponding row of data will be ignored by the platform.

Conversion Push

This aspect often does not receive the attention it deserves: in order to optimise the management of the Consent Mode, it is essential to also work on the Consent Management Platform. An inadequate configuration of these elements can have repercussions not only on the target audience, but also on conversions themselves. In a context where advertising campaigns are increasingly driven by artificial intelligence and depend on conversions for the nurturing and training of algorithms, poor campaign performance can be directly attributed to sub-optimal consent management.

Google has introduced two solutions to address these challenges. The first, which we have already seen, is Consent Mode, which shapes conversions to suit cases where user consent is not available. The second solution is Enhanced Conversion, a technology that allows a conversion signal to be sent even when the user is not tracked through traditional pixels, exploiting personal data. This technology can be implemented via APIs, allowing operators to use data collected in their Cloud Data Warehouse, such as the user’s email address, conversion date and transaction value, to transmit this information to advertising platforms. Since users often access various services using the same email, platforms are able to link the display of an ad to a specific conversion, thus attributing credit to the appropriate ad campaign.

This attribution process, once predominant in industries such as automotive – where the purchase may occur months after the first contact, following a test drive – is now considered essential for all types of business. Without this practice, a significant number of conversions may be lost. Modelling conversions through Consent Mode alone may not be effective enough.