With the announcement of the end of “third-party cookies” in the Chrome browser next year, digital advertising business models are in full upheaval. In this context, the CNIL commissioned an economic study of the possible consequences of this development and presents the main conclusions.

Digital advertising will tomorrow, even more than today, be at the heart of French media financing: according to a recent study commissioned by Arcom, digital advertising will represent 65% of the advertising market by 2030. In the At the same time, this market is affected by significant upheavals: deployment of the ATT system (App Tracking Transparency in English) in iOS, scheduled end of third-party cookies in Chrome planned for early 2025, rise of models of “consent or pay” cases, etc.

In this context, what will be the advertising business models of tomorrow? What role will alternative models play to the dominant solutions? More generally, what are the risks that these developments pose for data protection?

To answer these questions and try to anticipate developments, the CNIL asked two researchers from Télécom Paris, Christelle Aubert-Hassouni and Patrick Waelbroeck, to carry out an economic and competitive study on alternative digital advertising models to the dominant solutions.

The study was carried out at the end of 2023-beginning of 2024 based on 25 interviews with advertisers, media agencies, regulators and specialists in the digital advertising sector. It essentially aimed to better understand the market and new solutions, assess the privacy impact of new business models, but also to shed light on possible rebalancing of the advertising value chain.

Two questions motivated the study: will the current changes allow better respect for the privacy of Internet users? Will they work in favor of publishers?

Return to the role of third-party cookies

The first observation suggests that we should not overestimate the effectiveness of the targeted advertising model using third-party cookies (targeting, capping of exposure frequency, audience measurement). However, this model is popular with advertisers: the absence of third-party cookies, in certain current architectures, seems to result in a loss of value of auctions and therefore, a loss of value for publishers, with whom the large digital platforms are in a strong position to pass on price changes in the downstream market.

The study then notes that the entry into application of the GDPR did not result in a reduction in advertising tracking of users: the rates of non-consent, when it is required, having remained modest at least until the implementation of the ATT solution by Apple in April 2021. The upheavals in the sector are rather due to developments driven by private players themselves, in particular the large platforms (ATT solution or end of third-party cookies announced by Google).

Some recent research highlights that the deletion of third-party cookies when using the Chrome internet browser will result in strong competitive and organizational changes in the sector. From this point of view, the insecurity, often alleged, would be more economic and due to the behavior of the biggest players in the chain, than legal insecurity due to regulators.

Seven groups of solutions

Anticipating the removal of third-party cookies, researchers, after analyzing the available offer, have uncovered seven types of advertising solutions: the “privacy sandbox”, which allows targeting of users by cohort and by interests from navigation data, is described by the sector as the essential solution of tomorrow, and six models “alternative to the dominant solutions” (Google/Meta).

Three are alternatives using or generating other types of data:

• substitution identifiers (deterministic or probabilistic);
• contextual targeting (an old technique based in particular on keywords and reinforced using techniques in particular natural language processing); And
• cohort targeting, the purpose of which is to create audience segments.

Three are alternatives based on architectural choices:

• “retail media” which corresponds to all the advertising spaces offered by a distributor (traditional or online);
• environments traced by a user account – these two models relying essentially on proprietary or “first party” data; And
• toll tracker walls, in reality corresponding to paid business models in order to generate additional income.

The study analyzes these business models with regard to several criteria: technical integration, acceptability for the Internet user, ability to meet the needs of advertisers, merits in terms of protection of privacy and economic sustainability along the chain of value.

Ultimately, none of these solutions will be sufficient on their own to perform the same functions as third-party cookies in terms of targeting, measurement of frequency of exposure or audience.

More precisely, two major trends can be detected:

• The developments will favor players who will have a significant source of proprietary data such as distributors or identified mode universes (often described under the term “closed gardens”), currently not always collected for advertising purposes.
• For open Internet players, who have less access to proprietary data, it will be necessary to implement several complementary solutions in order to cover the entire inventory and functionalities, leading to technical complexity and a challenge of interoperability.

New players, business models and competitive dynamics

Another observation of the study is that these developments generate strong competitive issues, for three reasons according to the researchers. First, large platforms tend to define privacy protection as a ban on tracking Internet users outside of their closed universes. Second, some of these players benefit from strong vertical integration in the advertising addressing chain. Thirdly, the multiplication of types of terminals calls into question their interoperability, with the unequal ability of players to respond to these challenges.

In this context, the market benefits from new entrants (such as Internet service providers) and value will be captured tomorrow by a greater number of solutions than today, none of them having the answer to everything , not even the privacy sandbox, whose economic model is still largely undefined (see PeREN study). On the other hand, the competitive advantage will go more to closed architectures and retail media, which promises to be the big beneficiary of the current developments.

The scenario favored by researchers is not “the end of the open Internet” and the closure of the digital economy around a few large players who alone would have the data necessary to sell digital advertising. Indeed, there is a relationship of mutual dependence between the major players and those of the open internet: the former needing the latter in terms of content creation. But outside of “closed gardens”, the constraints of sharing and reconciling data will be more demanding than yesterday, without us knowing today what type of actor will be required tomorrow to carry out these operations.

In this context of the growing role of proprietary data but also of operational complexity, and therefore increased costs, the study concludes that large publishers will be in a better position than small ones, who will have to be active, pool their data to reach a mass critically, through cooperation or mergers, and being vigilant about interoperability issues to make wise investments.

Close cooperation between the CNIL and the Competition Authority therefore remains necessary to observe developments in this advertising market, identify risks to privacy and competition, and implement regulatory synergies. Their common approach was also published in December 2023 in the joint declaration.

A renewed role of data

While the introduction of the privacy sandbox was presented as a desire to improve privacy, future developments will not, according to researchers, lead towards a significant reduction in tracking or a reduced role for personal data on the advertising market.

This role will, however, be renewed by the use of new types of (proprietary) data, new channels for sharing this data (for those who need it and those who have an interest in giving it away) and new synchronization methods for find good performance in terms of individual attribution.

The main point of attention that the CNIL draws from this study is a phenomenon of opening and reuse of purchasing data. Until now, distributors had no interest in sharing them or reusing customer loyalty histories for advertising purposes. This is changing with the rise of retail media, and other financial players could in turn market advertising targeting. Identity reconciliation is a major issue. The CNIL will be particularly vigilant to ensure compliance with these developments.

Furthermore, researchers share a certain skepticism in the sector regarding so-called “consent-free” solutions. On the one hand, contextual type solutions are not necessarily synonymous with absence of consent, because they may involve the processing of personal data, for example for audience measurement, although in smaller quantities. On the other hand, they are not very popular with advertisers and are probably dominated by other solutions on the market. The most virtuous alternative in terms of data protection, they would however require significant voluntarism from the European regulator to be able to strengthen their interest for the sector.

A market which has not yet identified clear choices

To conclude, the interviews carried out by the researchers paint the image of a sector that is partly wait-and-see, with the emergence of multiple technical solutions whose respective merits have not yet been tested by the natural play of the market. Some solutions are still in the testing phase, some more or less conclusive, and are likely to evolve. This state of affairs was further reinforced by Google’s recent announcement to postpone the end of third-party cookies in Chrome from July 1, 2024 to January 1, 2025.

Finally, the methodological limits of the study cannot be ignored: it only concerns French players, provides a snapshot of the market at the end of 2023-beginning of 2024 and did not investigate new types of terminals (mobile applications, Connected TV) which make the economic dynamics of the digital advertising market in France even more complex today.

An economic approach to reduce uncertainty

In conclusion, at the end of this prospective exploration focused on online advertising business models, it turns out that current uncertainties are generated at least as much by the economic strategies of major digital players, who have their own conception of privacy and their own value propositions in this area, as well as through regulatory developments.

The rule of consent thus seems to be well integrated by the market and adaptable to a variety of competing solutions. As for Internet users, the CNIL’s evaluation of its action plan on cookies via surveys showed that the rate of refusal of targeted advertising on the French Internet had stabilized at less than 40% in June 2022, and did not reveal any particular phenomena of “consent fatigue”.

This observation of regulatory stabilization should be seen in light of very recent developments moving in the direction of the development of “consent or pay” type models on the social media market. The sector is questioning the relevance of this model in the field of media and press, an area in which it has already developed in Germany, for example.

This study on alternative advertising models usefully informs the CNIL’s thinking in its regulation of targeted advertising and demonstrates its desire to dialogue with market players. This approach should make it possible to produce ever more appropriate flexible law and support tools.

https://www.cnil.fr/fr/publicite-en-ligne-la-cnil-se-prepare-aux-evolutions-des-modeles-daffaires