Google gives publishers leverage against AI traffic

Google seeks to address a distribution crisis that extends beyond publishers

Google is introducing a new feature intended to give greater weight to users' explicit choices in how they receive information. As TechCrunch reported in its article titled Google gives publishers a new way to fight AI-driven traffic losses, users can now set a publisher as a preferred source. This preference may improve that outlet's presence across several areas of the Google ecosystem, including Google Search, Discover and Google News.

At first glance, the change may seem modest: a button, a preference, a more direct relationship between a reader and a news outlet. But it comes in a much broader context. News publishers are facing a weakening of their traditional role in the digital distribution chain. For years, search engines have served as an essential gateway to articles, investigations, analyses and breaking news. Platforms were not neutral, and their algorithms have always determined part of editorial visibility. Nevertheless, they remained major audience providers for many news outlets.

The rise of answers generated by artificial intelligence has changed the nature of this relationship. When a search engine directly answers a question, summarizes several sources or synthesizes information within its own interface, users can obtain an answer without opening the pages that helped feed it. The phenomenon of zero-click searches is not new, but conversational interfaces and AI-enhanced answers give it an additional dimension: they no longer merely display links; they attempt to handle the request before the user has to choose a site.

For publishers, the stakes are not merely symbolic. A visit potentially represents advertising revenue, a subscription conversion, a newsletter signup, exposure to a brand or the opportunity to build a lasting relationship with a reader. Conversely, an answer consumed entirely within the search engine's environment reduces opportunities for monetization and retention. The risk is particularly acute for publications whose model relies heavily on search-driven audiences, whether they cover general news, operate specialized sites, serve local communities or invest in practical and explanatory content.

The feature announced by Google therefore does not merely change how results are displayed. It places a question at the center of news distribution: can a reader regain some control by explicitly indicating which publishers they want to see more of? TechCrunch presents this initiative as a tool enabling publishers to fight AI-related traffic losses. However, this wording deserves to be examined closely. A user preference can strengthen a loyalty signal, but it does not guarantee an audience, does not replace a clickable link and does not, on its own, resolve the question of the value captured by platforms when they answer directly.

The development also fits into Google's particular history with news. The group has long occupied a structuring position in access to online content through its search engine, Google News and Discover. For news outlets, this centrality has created an ambivalent dependency. Google can highlight a publication, facilitate discovery of its articles and send it a large volume of visitors. But any change to interfaces, ranking criteria or product policies can also abruptly alter the conditions of access to an audience.

In this context, the ability to declare a favorite news outlet has implications that go beyond the feature itself. It implicitly recognizes that the relationship between an internet user and a publication should not be entirely inferred by recommendation systems. Google already has many signals to personalize its products: browsing history, queries, interactions, context and current events. The button adds a clearer, more voluntary and potentially easier-to-understand signal: the reader indicates that they want to receive more content from a given publisher.

What the preference granted to publishers changes in practical terms

According to information reported by TechCrunch, Google allows users to choose a publisher as a preferred source through a dedicated button. Once this choice has been made, the preference can help increase that outlet's visibility in Google Search, Discover and Google News. These three surfaces are important, but they do not serve the same use case: search generally comes into play when a user enters a query; Discover offers a personalized selection of content; Google News organizes access to news according to editorial, topical and personalization criteria.

The first notable point is that Google does not present this setting as a ranking guarantee. A preferred source may be more visible, which differs from a promise of systematic appearance, top placement or a specified volume of traffic. This nuance is fundamental. In search, the answer to a query must continue to depend on factors such as the relevance of the content, its timeliness and the user's context. In Discover and Google News, the reader's preference is added to selection mechanisms that remain controlled by Google's products.

The second point is the explicit nature of the signal. Algorithms usually attempt to infer users' interests from their behavior. Yet reading an article, performing a search or clicking a link does not necessarily mean that a reader wants to receive content from the same publication regularly. Conversely, a clearly stated preference creates a clearer relationship. For a news outlet, this relationship resembles, in principle, other forms of consent to receive attention: subscribing to a newsletter, following an account on a social network, adding a feed to an aggregator or enabling notifications.

This comparison has its limits. A preferred source in Google does not give the publisher direct access to the user's contact details or control over the frequency of distribution. The outlet remains present in a third-party environment governed by Google's interfaces and rules. It is therefore not an owned channel in the way a newsletter subscriber base can be. But it is potentially a way to make an editorial preference visible at one of the web's largest information gateways.

The measure could be useful to publications that already have a strong identity. A reader who trusts a daily newspaper, a local news outlet, a scientific journal or a specialized site can now give concrete form to that preference instead of relying solely on recommendation mechanisms. For publishers, the challenge is then to convince users that this action is worth taking. The button does not create loyalty; it gives a technical expression to pre-existing loyalty, or to the desire to build it.

This dimension is particularly important at a time when news content is increasingly consumed in the form of summaries, excerpts, notifications or short answers. In these formats, the editorial brand can become less visible than the subject itself. A user seeking an explanation of a political decision, a technological innovation or an international event may settle for an overview supplied by a platform without remembering the name of the source. The preference mechanism puts, at least in part, the publisher's name back at the center of the journey.

It is also necessary to distinguish the feature from direct access to a site. When a user types a news outlet's URL, uses its app, opens its newsletter or subscribes to its alerts, the relationship is more immediate. The preference in Google operates at the stage of intermediation: it attempts to improve the chances that an article will be found in an environment where many publications, rich results and generated content compete for attention. Its value rests precisely on this intermediary position, but this is also what limits its autonomy.

In the face of AI answers, reader choice becomes a strategic but incomplete signal

The announcement must be viewed in light of the transformation of the search engine by artificial intelligence. Google has progressively enhanced its results with formats capable of answering queries more directly. The group notably introduced AI Overviews, which aim to provide synthesized answers for certain informational needs. For users, the goal is clear: speed up access to useful information. For the sites that produce that information, the effect may be more ambiguous if the answer reduces the need to visit source pages.

The problem is not that every direct answer would mechanically eliminate clicks. Some searches still require a full reading, context, verification, expertise or a decision that a few lines cannot provide. Investigative content, in-depth analyses, detailed comparisons, local data and articles with high editorial value retain their own role. But publishers' economic balance may be weakened when the simplest, most frequent or most easily synthesized queries stop bringing in visitors.

In this situation, the user preference offers a response different from one consisting solely of negotiating with platforms or blocking access to content. It seeks to restore a form of active demand. If a reader indicates that they want to follow a publisher, Google has an additional reason to offer that outlet when its content is relevant. The mechanism can be seen as an attempt to shift recommendation from a purely implicit logic to a partially declarative one.

This is an important point because generative AI systems raise a visibility issue as much as a compensation issue. A synthesized answer may cite, link to or mention sources. But the presence of a citation guarantees neither that the user will see it, nor that they will understand it as an invitation to consult the article, nor that they will click it. In an environment where the interface seeks to satisfy the request immediately, preference for a publication can serve as a reminder: beyond the generic answer, the reader also wants to receive that source's coverage.

However, it would be excessive to view it as an automatic remedy for traffic losses. The first limitation is adoption. A button has an effect only if users discover it, understand its usefulness and take the time to select publications. In digital uses, every additional step reduces the number of users who complete an action. The best-known outlets or those closest to their communities may benefit more easily than small newsrooms, newcomers and publishers whose brand is less established.

The second limitation concerns the actual weight given to the signal in ranking and recommendation systems. Google says that the preference can improve a source's visibility. But the extent of this improvement will depend on rollout arrangements and product trade-offs. A preference may count heavily in some situations, less in others, or be offset by the nature of the query, current events, source diversity or other relevance signals. Without precise visibility into these mechanisms, publishers cannot turn the announcement into a traffic forecast.

The third limitation lies in the scarcity of attention. Even if a user follows several news outlets, they will not be able to read every publication offered to them. In Discover, Google News or search results, the available space is limited. Favoring already preferred sources may help retain existing audiences, but it does not necessarily resolve the challenge of discovering new readers. Yet that discovery is essential for publishers seeking to grow their audience or offset the erosion of certain channels.

The feature may also raise a question of informational balance. Personalization makes it possible to bring readers closer to outlets they trust, which can strengthen reading habits and support identified newsrooms. But it may also increase the share of content coming from a limited number of sources if the user does not diversify their preferences. News products must therefore reconcile recognition of individual choices with diversity of viewpoints, territories and editorial treatment.

A retention tool in an already saturated competitive landscape

Google's proposal comes as news outlets have long been trying to reduce their dependence on major platforms. Newsletters, mobile apps, podcasts, subscriptions, events, notifications and direct access are all ways of maintaining a connection with readers without letting an intermediary alone define distribution. The preferred source replaces none of these channels. At best, it can complement a strategy in which the outlet works simultaneously on its brand, search visibility, community and subscription products.

The comparison with social networks is instructive. On these platforms, publishers have often faced changes in priorities: a modification to the algorithm, a new format or an interface change can sharply reduce organic reach. Following an account has never been an absolute assurance of distribution to all followers. Likewise, being selected as a preferred source in Google should not be interpreted as an automatic right to appear throughout all of a user's information journeys.

The difference is that Google frequently operates at the moment of intent. An internet user conducting a news search or consulting Google News expresses a more direct informational need than someone scrolling through a social feed. This position can make the preference signal particularly valuable: it links a news outlet to a moment when the reader is actively seeking an answer or has opened themselves to the discovery of information. In Discover, the stakes are somewhat different, since the interface relies on anticipated recommendations of the user's potential interests.

For publishers, the operational challenge will be to turn a technical feature into a reader-driven action. They will need to explain simply what a preferred source is, without promising visibility they do not control. Communication that is too aggressive could create mistrust, especially if the user does not quickly perceive a difference in results or recommendations. Conversely, a clear editorial message can present this setting as a way to support journalism one already consults, alongside subscribing, receiving a newsletter or visiting the site regularly.

Major brand-name news outlets have an obvious advantage: they are more easily identifiable and can mobilize their existing audiences through their own channels. But the tool is not reserved, in principle, for the most powerful groups. A recognized regional title, trade publication or specialized outlet can also have a highly engaged community. In some areas, value does not come from a massive volume of occasional readers, but from a smaller audience that regularly seeks the same expertise. Explicit preference could have particular usefulness there.

On the other hand, the measure does not directly address the discoverability of emerging sources. If users mostly select names they already know, new outlets will still have to earn their place through the quality of their coverage, their presence in results, their search visibility and their ability to build a reputation. Google's algorithms therefore retain a major responsibility in exposing content from less established publications, local newsrooms and specialized voices.

The announcement also highlights a deeper tension: Google is asking users to play a more active role in defining their information diet, even though major platforms have accustomed the public precisely to highly automated consumption. Many internet users do not consciously organize their sources; they follow the links presented, topics that emerge or recommended content. The success of the feature will depend on Google's ability to make it understandable and accessible, but also on readers' willingness not to delegate all their choices to algorithmic recommendation.

Specific implications for French and European media

For French publishers, the debate over visibility in search engines and AI interfaces cannot be separated from the debate over the value of journalistic content. France and the European Union have placed press neighboring rights at the center of discussions between platforms and publishers. These rights rest on the idea that companies that reuse or derive value from press content must recognize newsrooms' economic contribution. Source preference does not directly address this issue: it concerns distribution and visibility, not compensation for the use of content.

This distinction is essential. A news outlet may receive more readers thanks to better exposure while still facing compensation issues for the use of its content by search or AI products. Conversely, a compensation agreement does not guarantee that users will see or visit more of the articles concerned. Publishers must therefore contend with two complementary dimensions: obtaining fair recognition of the value created by their work, and maintaining distribution channels capable of turning that value into a relationship with the public.

The French-speaking market has several specific features. It includes national, regional, local, business, cultural, scientific and specialized publishers, with very different economic models. Some have developed a strong subscription culture; others rely more heavily on advertising, local distribution or associated services. The impact of a preference in Google will therefore not be uniform. A local-interest outlet may hope to strengthen its connection with readers attached to their area. A specialized publication may see it as a way to remain present for a professional audience. A general-interest title will probably seek to increase how often its regular readers return.

The language issue also matters. Search and recommendation systems have long been optimized in contexts dominated by English, even though Google operates in many languages. For French-speaking readers, the ability to favor a known publication may be a way to better steer selection toward sources that directly cover French, Belgian, Swiss, Quebec, African or European issues in French. This does not eliminate visibility difficulties, but it theoretically reduces the share of uncertainty over the editorial origin of certain recommended content.

The tool's actual reach will nevertheless depend on its accessibility across countries, languages and products. Google's announcements are often rolled out gradually, and features may vary across markets. For French newsrooms, it would be premature to build a strategy based on a general promise as long as the precise conditions for access, configuration and visibility are not known in their environment. The issue is also practical: a tool that is not very visible or is limited to a fraction of users cannot become, in the short term, a pillar of distribution.

European media will also need to monitor the transparency Google provides regarding the effects of this preference. The most useful information is not merely that relating to the number of outlets available in the feature, but that which makes it possible to assess its effectiveness: what share of users uses it, in what contexts is a preferred source given greater prominence, and how does this choice translate into impressions, clicks and visits? Without sufficiently precise measurement tools, publishers risk facing a signal whose existence they know but whose real value they cannot assess.

This measurement requirement is particularly important for small players. A large newsroom can experiment with several channels, invest in audience teams and absorb a period of uncertainty. A local or independent publisher often has fewer resources to adapt its work to every platform change. If source preference becomes a significant lever, its operation will need to be sufficiently clear so that it does not favor only those that can afford to run campaigns to mobilize their audience.

Explicit preference can become a counterweight, not a substitute solution

The scope of this announcement will depend less on the existence of the button than on the behavior it manages to establish. If users get into the habit of choosing news outlets they consider reliable, useful or close to their interests, Google could more visibly incorporate this logic of human selection into its news products. This would offer publishers an additional lever at a time when AI interfaces risk making sources less central to the search experience.

The favorable scenario is one in which preference creates a virtuous cycle. A reader selects a news outlet; they see its content more often when it is relevant; they learn to recognize its coverage; they return more regularly; and this relationship may eventually lead to a subscription or another direct connection. Under this hypothesis, the feature would not mechanically restore traffic lost to generated answers, but it could help protect some of the strongest editorial relationships against the gradual disappearance of brands in interfaces.

The less favorable scenario is equally plausible. Users may ignore the setting, select very few sources or notice no visible change after making their choice. Google may also give this signal limited weight relative to other relevance and personalization criteria. In that case, the feature would be useful as a gesture of recognition toward publishers, but its economic impact would remain marginal. TechCrunch emphasizes precisely that effectiveness will depend on adoption by readers and the actual place given to these preferences in algorithms.

The real test will therefore be transparency. For the tool to be credible, publishers must be able to understand, at least in broad terms, what it changes in their visibility. Readers must know what they get by choosing a source and retain the ability to modify their preferences easily. Google, for its part, will need to reconcile this promise of individual control with the requirements of quality, diversity and relevance that structure news products.

Over the longer term, the announcement reveals a shift in the relationship between search engines, AI and the press. For a long time, the debate focused on how platforms rank content. With generative answers, it increasingly concerns whether content remains accessible as a destination or becomes invisible raw material for an answer produced elsewhere. Source preference does not address all dimensions of this shift, particularly those related to attribution, clicks and compensation. It nevertheless reintroduces an important idea: readers may have an interest in actively choosing who informs them.

For French-speaking publishers, the most robust strategy will probably remain one that depends on no single intermediary. Google's feature can become a tool to integrate into a broader policy of retention, editorial quality and direct relationships. But its value will not be measured by the number of activated buttons: it will be measured by its ability to maintain readers' access to the articles themselves, at a time when search engines are evolving toward interfaces capable of answering before even sending a click.