Google Preferred Sources: What Are They and How Do You Apply?

Google now lets users choose whose content they want to hear. Since May 2026, each user’s list of favorite websites has influenced what appears in the responses generated by artificial intelligence, with a visible label.

For content creators, the question changes. Instead of asking how to rank higher, it’s worth asking who already trusts you enough to choose your domain.

That’s where Google Preferred Sources comes in—the feature that turns stated preferences into recurring visibility.

The difference is significant because almost all other search efforts are competitive. Here, there’s no auction and no ranking to win—just a request to make to people who already read your content.

What are Google Preferred Sources, and how does the feature work?

"Google Preferred Sources" is a feature that allows users to mark websites as favorites within Search.

Based on this selection, Google begins to highlight content from those domains for that user, with a “preferred source” label in Top Stories and in AI-generated answers.

The feature originated in the News tab and expanded from there. Google’s documentation describes the logic straightforwardly: when a user selects a website as a preferred source, its content is more likely to appear in Top Stories with the “preferred” label.

There is a technical limitation that affects the implementation: eligibility is determined by domain and subdomain. Google specifies that addresses such as https://www.exemplo.com/ and https://codigo.exemplo.com/ are valid, but https://www.exemplo.com/blog is not.

In practice, when someone chooses your site, they’re choosing the entire site, not just a specific folder. The average quality of the domain becomes a factor; for example, someone who adds an e-commerce site because of its buying guides will also see their category pages and corporate announcements featured.

Adoption is growing rapidly, based on figures released by Google itself. On May 27, 2026, the company reported more than 345,000 selected unique sources. On August 20, 2026, the reported number had already surpassed 600,000.

It’s worth noting what this data is—and what it isn’t. These are figures self-reported by the platform, with no published methodology, and they serve to gauge adoption, not to predict the performance of a specific website.

Caption: The Google Preferred Sources feature allows readers to add your website as a favorite source, highlighting its content in search results and AI responses.

Why do preferred sources mean visibility gained through preference?

Preferred sources are visibility gained through preference because the trigger isn’t the algorithm—it’s a person. No one optimizes a page to be chosen as a favorite: the reader decides, clicks, and that choice takes effect in their future searches. The leverage shifts from the code to the relationship.

This shift reshapes the strategy. In traditional SEO, the goal is to improve the content and hope the system recognizes it. With preferred sources, the goal is to ask—in the right place—those who are already reading.

Google states in its May 2026 announcement that “people are twice as likely to click on a preferred source.” The statement comes from the company itself, with no methodology disclosed, so treat it as a directional indicator, not as a multiplier to be included in a projection spreadsheet.

Still, the direction makes sense given how it works, since anyone who specifically asked to see that site has a clear intention. It’s the opposite of a cold click from someone who just happened to come across it in the SERP.

There’s a second, less obvious effect on search behavior with AI. Since a large portion of search journeys end without a click—as described in “zero-click search”, the “preferred source” label provides a concrete reason for the reader to leave the search result and go to the page.

None of this replaces the fundamentals. The work of answer engine optimization remains what makes content searchable and citable; preferred sources come into play afterward, increasing the likelihood that your material will be the one cited by that audience.

There is, therefore, a clear division of roles. Well-structured content addresses technical eligibility, and the invitation addresses human preference.

Where does the preferred source appear in AI Overviews, AI Mode, and Top Stories?

The preferred source currently appears on five Search surfaces, all with visible indicators showing who made the selection.

The turning point came in May 2026, when Google integrated the Top Stories feature into its AI experiences and created carousels that also highlight the selected sources.

Each surface provides a different type of exposure, and it’s worth understanding which one resonates best with your content:

Table: Surfaces where the user’s selection changes what they see, according to Google’s May 2026 announcement and publisher documentation.

Two practical observations emerge from this overview. The first is that Top Stories remains the most direct gateway to news content, and Google reports that the feature is available globally on this platform, in the languages where Search operates.

The second is that the highlight within the AI-generated overview and within the Google’s AI mode applies to evergreen content as well, not just to news.

This opens the door for those who don’t publish journalism. Comparative documentation for a SaaS platform, a technical specification guide for an industry, or guidance material from a clinic could be the cited excerpt and be labeled as a preferred source.

In the same May update, Google expanded the “highly cited” labels, which indicate when an article references sources that are frequently cited by other outlets. It’s an additional signal, useful for those who produce their own investigative reporting and original content.

How can you include the “preferred source” invitation on your website, blog, and newsletter?

Adding the “preferred source” invitation takes just two steps: choose the method for adding it and choose where it appears.

The method recommended by Google is a button embedded directly on the page; the deep link to Search preferences is the alternative for those who cannot run the script.

In the recommended approach, the documentation instructs you to load a script into the page’s`<head>` and place a container where the button should appear. The button supports theme and language customization, which eliminates the need for workarounds to adjust its appearance.

On August 20, 2026, this button became more user-friendly. Google now directs the reader back to the exact spot on the page where they were after confirmation, rather than sending them to a preferences screen and leaving them there.

When embedding a script isn’t possible, use a deep link. The documented format is https://www.Google.com/preferences/source?q=exemplo.com—replace the example domain with your own—and it opens directly to the add screen for that site.

Once the mechanism is in place, what remains is the part that drives results: distribution. A button hidden in the footer yields almost nothing; the invitation needs to appear where the person has just received value.

Here’s how this breaks down by channel:

Table: Distribution map of invitations by channel, with recommended format and suggested frequency of repetition.

The invitation text matters more than the button design. It works best when it explains the benefit to the reader, not to the brand: saying that they’ll find the content more easily in their searches is more honest and more effective than simply asking for support.

One final recommendation regarding the sequence: Start with the blog and the newsletter, which have the most engaged audience, and only then move on to broader channels, where the sign-up rate tends to be lower.

What advantage does having your own reader base provide in terms of “preferred sources”?

Having your own reader base is the biggest advantage in “preferred sources,” because access is granted on a request basis rather than through an auction.

Those who already have an ongoing relationship with an audience have people to reach out to, but those who rely solely on cold search traffic need to win over the reader and secure an invitation during the same visit.

The key factor here isn’t audience size—it’s contact frequency. Newsletter subscribers, client portfolios, communities on messaging apps, engaged followers, and a B2B base with active contracts all share the same characteristic: they already recognize the domain and open whatever comes from it.

Comparing this to paid media helps put the benefit into perspective. A brand that only buys attention must build from scratch the reason to be chosen as a source of information; a brand that already publishes regularly simply needs to highlight the benefit and show where to click.

Each type of audience responds to a different pitch, and the invitation is most effective when it respects that. E-commerce repeat customers respond to practical utility, SaaS users to product updates, clinic patients to reliable guidance, and B2B buyers to market insights.

The seasonality of the business determines the optimal window for making a request. Peaks in informational searches—such as the day before Black Friday in retail, quarter-end in B2B, peak season in tourism, or annual renewals for professional services—concentrate spontaneous attention and reduce the cost of the invitation.

One important consideration rounds out this reasoning. A channel for engaging paying customers must not become a channel for pressure: the invitation should be presented as a convenience, with a clear explanation of how it relates to that person’s search and how they can opt out.

How do you measure “preferred sources” in the generative AI report in Search Console?

The Search Console generative AI report is where the effect begins to show up in the data, albeit partially. Google announced the reports in June 2026, and they show how a website’s content performs within Search’s generative experiences.

The scope calls for honesty about what exists and what does not. According to specialized coverage of the launch, the report includes impressions, pages, countries, devices, and dates, and treats AI Overviews and AI Mode together, without separating the interfaces.

Two limitations change how the report is interpreted. According to Search Engine Land, the report does not provide click data from AI responses, and the launch was initially limited to a subset of websites, with gradual expansion.

In other words: don’t expect a metric called “preferred source clicks.” No Google platform currently reports this, and promising this metric internally sets expectations that the tool cannot meet.

The viable approach is to build an approximate dashboard. Combine the trend in AI impressions in Search Console, detailed in the AI report in GSC, with behavioral signals that you can control.

Three approaches support this analysis without inventing metrics:

1. Clicks on the invitation itself: how many people clicked the button or the deep link, measured by the event on the website and by the click in the email.
2. Direct and brand traffic: growth in brand searches and direct traffic during the same period as the sign-up campaign.
3. Reader retention: the proportion of returning visitors, which is the expected behavior of those who subscribed.

None of the three metrics proves causality on its own. Together, and when compared to the period before the campaign, they indicate whether the invitation is gaining traction or falling flat on the page.

It’s also worth aligning the report’s terminology with the rest of the content operation. An AI impression is not a visit, and this distinction must be clarified before the figure is presented at a results meeting.

What should you avoid doing to stay within Google’s spam policies?

The golden rule is simple: the invitation may be persistent in its distribution, but it must never be coercive or deceptive.

Google’s Google’s spam policies define spam as a technique to deceive users or manipulate search systems, and the official text explicitly includes attempts to manipulate generative responses.

This became clearer in June 2026, alongside that month’s spam update, when Google reiterated that its existing policies apply to attempts to manipulate AI-generated responses.

There was no new policy regarding preferred sources, but rather confirmation that the existing guidelines cover this new territory.

With that in mind, certain practices are out of the question. Each item below has a specific reason behind it; this is not just legal nitpicking:

1. Making access or benefits conditional on following: requiring a person to follow the site to unlock content, enter a giveaway, receive a discount, or qualify for free shipping constitutes the purchase of preference, not preference itself.
2. Automating or simulating the follow: any script that clicks on the user’s behalf, or that triggers the follow without the user’s conscious action, constitutes system manipulation.
3. Hiding what the button does: Labeling the prompt as “accept” or “continue” falls under the category of misleading functionality.
4. Inflating volume to dominate search results: Mass-posting shallow content on the domain to appear more frequently in search results is exactly the kind of large-scale content abuse described in the policies.
5. Promising a ranking boost: Telling readers, customers, or management that joining improves the site’s ranking on Google is a false claim.

There is also a reputational risk that no policy addresses. Asking for sign-ups across every channel, every week, wears down the user base and burns the invitation precisely with those who were most likely to accept it.

The antidote is the same as always: content that deserves to be chosen. It’s worth reviewing the domain’s credibility and editorial consistency before spending energy on a follow campaign, because the choice affects the entire website.

How does this change strategies for educational marketing?

In strategies for educational marketing, “preferred sources” have little impact on the technique but a significant impact on the timeline.

The sector has a distinct seasonal cycle and involves high-consideration decisions, so the invitation has better time windows and targets an audience that is already visiting the domain out of necessity. This serves as an example of the sector, not as a rule exclusive to it.

This is a sector-specific snapshot, and the logic applies to any niche with these two characteristics. Real estate, healthcare, and B2B software all follow the same pattern: a lengthy purchasing process, extensive research before initial contact, and an audience that returns to the website multiple times.

Seasonality is the first key difference. Periods such as the opening of registration, the release of results, and the start of the new semester see a surge in spontaneous informational searches, and these are the windows during which the enrollment request faces the least competition.

The second point is the nature of the audience. Students, applicants, faculty, and alumni already receive communications from the brand for other reasons, which makes the invitation plausible without coming across as a sales pitch.

The third point is where the evidence itself resides. The program page is the resource that brings together the curriculum, faculty, and admission criteria, and it is precisely the type of primary content that supports citations in AI-generated responses.

There is a specific risk that warrants attention. Much of the content circulating in the sector comes from partners or third parties, and partner content published on the institution’s own domain is factored into the average quality score selected by the reader.

Local relevance rounds out the list. Educational marketing strategies for a regional institution benefit more from targeting the city and its surrounding area than from a diluted national reach, because the decision is geographically driven.

Frequently Asked Questions About Google Preferred Sources

So, is it worth applying to become a Google Preferred Source?

Yes, and for a reason that’s rare in search: the outcome depends on a decision you can directly influence. No algorithm update can take away your ability to apply, and no competitor can outbid you for the spot.

The work is less glamorous than it seems. It boils down to placing the invitation in the right places, writing an honest explanation, repeating it sparingly, and tracking AI impressions along with your own behavioral signals.

The risk of being left out, on the other hand, is silent. As the number of selected sources grows, those who don’t apply simply don’t make it onto anyone’s list and lose their branded presence on platforms that are becoming the standard for reading.

If your organization already produces consistent content and wants to turn that output into a stable presence in AI responses, the path involves a review of structure and distribution. It’s this kind of assessment that the mkt4edu’s SEO team conducts—from technical inventory to content planning.

Those who are still deciding how much of their budget to allocate to search can find the other side of the equation in whether investing in SEO is worth it, with the ROI data that supports the decision before moving forward with the proposal.

To discuss your company’s situation and figure out where to start, talk to the mkt4edu team.