On May 27, 2026, Duncan Osborn, Product Manager for Google Search, announced that Preferred Sources — a personalization feature that had existed since a June 2025 Labs experiment and reached full global language availability on April 30, 2026 — now extends into AI Overviews and AI Mode. Within days, Google reported more than 345,000 unique sources had already been selected by users, up from roughly 200,000 at the December 2025 global expansion, and stated that users are twice as likely to click through to a source they have personally marked as preferred compared to an unmarked source in the same position.
The mechanic is simple to describe and easy to underestimate. A reader visits their Google Search personalization settings, or clicks a star icon next to Top Stories, and adds a specific publisher by name or URL. From that point, when that publisher's content appears inside a generated AI Overview or AI Mode response for that specific user, the link carries a visible "Preferred" badge. Google has also confirmed, in the same announcement window, that this is not purely a cosmetic label: preferred status is described as becoming an active input into how frequently that source gets cited in AI-generated answers, not just a badge applied after citation selection has already happened by other means.
Why Does This Matter More Now Than When Preferred Sources Only Applied to Top Stories?
Ahrefs' February 2026 analysis found that websites ranking first in classic organic search now receive 58% fewer clicks when Google displays an AI Overview above the results for that query. Chartbeat data published in March 2026 found small publishers specifically have lost 60% of their search-driven traffic over roughly two years, with larger publishers losing a smaller but still substantial 22% over the same period. Against that backdrop, Preferred Sources extending into the two surfaces now absorbing the largest share of user attention — AI Overviews and AI Mode, the latter of which crossed one billion monthly active users globally as of May 2026, according to Google's own I/O announcements — turns what was a Top Stories personalization nicety into one of the only mechanisms Google currently offers that puts a measure of visibility control back into a publisher's hands, rather than leaving citation entirely to algorithmic selection.
The important caveat, and one that search-industry coverage converged on quickly: this is a user-driven feature, not an algorithmic ranking lever a publisher can influence through content or technical work alone. A reader has to take a deliberate, personal action inside their own Google account settings. Google's own standard relevance and quality signals still apply on top of preferred status — being marked preferred does not override the underlying content still needing to be fresh and relevant enough to be a citation candidate in the first place.
What Is the Separate "Highly Cited" Badge, and How Does It Differ From Preferred?
Alongside the Preferred Sources expansion, Google is separately rolling out a "Highly Cited" label that identifies original reporting or research that many other outlets subsequently reference and cite. Unlike Preferred Sources, this badge is not user-selected — it is calculated automatically by Google based on cross-publication linking and citation patterns, rewarding primary source material specifically. If your publication runs an original study and publishes the findings first, you qualify for consideration. If another site subsequently writes up those same findings, referencing your original work, that secondary coverage does not independently qualify for the Highly Cited label, even if it ranks well or is itself well-written.
For a content strategy, the two badges reward different but genuinely complementary behaviours. Preferred Sources rewards audience loyalty and requires an existing, engaged readership willing to take a deliberate action. Highly Cited rewards genuine primary research and original reporting — precisely the non-commodity content principle covered in Google's own May 2026 generative AI optimization guide and in the companion article on non-commodity content elsewhere in this series. A publisher pursuing both simultaneously is building trust on two independent axes: one measuring how much your existing audience values you enough to name you specifically, and one measuring how much the broader publishing ecosystem treats your original work as the source worth citing.
What Are the Actual Eligibility Requirements for Preferred Sources?
Google's stated eligibility bar is deliberately broad and low: any website that publishes fresh content is eligible, with the sole structural restriction being that eligibility applies at the domain or subdomain level rather than at an individual blog path — meaning example.com or blog.example.com can qualify, but example.com/blog specifically cannot be nominated as a distinct preferred source separate from the domain it sits on. There is no exclusive approval process or manual review gate visible to publishers; a site's actual selectability is confirmed simply by checking whether it appears in Google's own source-preference search tool when a user searches for it.
How Does a Publisher Actually Prompt Readers to Add Them as a Preferred Source?
Google provides two concrete mechanisms specifically designed to make this ask straightforward to operationalise. The first is a direct deeplink, formatted as google.com/preferences/source?q=your-site-url, which takes a user immediately to the specific preference toggle for that publication rather than requiring them to search for it manually inside Google's settings. The second is a set of downloadable "Add as preferred source" button graphics, available in sixteen languages, explicitly designed to sit alongside existing social-follow calls to action on an article page or in an email newsletter footer.
The practical implementation sequence that follows from this: place the deeplink or button prominently in your highest-engagement channels first — email newsletter footers, post-article CTAs, and social bio links — since the audience most likely to take this deliberate action is the audience that already demonstrates loyalty through open rates, repeat visits, or existing social follows. A cold, undifferentiated site-wide banner asking every visitor to add you as a preferred source will convert at a meaningfully lower rate than a targeted ask placed where your most engaged readers already are.
Does Preferred Source Status Directly Change Citation Selection, or Only Citation Display?
This is the detail that separates the May 27 announcement from a purely cosmetic UI change, and it is worth stating precisely because several early industry summaries conflated the two. Google's own confirmation is that preferred status functions as a genuine input into how frequently a source gets cited in AI-generated answers for that specific user, not merely a label attached retroactively to citations that would have occurred regardless. The scale of that citation-frequency effect has not been quantified by Google, and standard relevance and quality signals continue to apply on top of it — a preferred source with genuinely low-quality or irrelevant content for a given query is not guaranteed a citation purely on the strength of the user's preference. But the direction is unambiguous: audience loyalty, expressed through this specific mechanism, now has a measurable connection to AI-generated answer visibility, which is a genuinely new category of signal distinct from the content-quality and technical-structure factors covered throughout the rest of this series.
What Does This Mean for a Non-News, B2B, or SaaS Content Programme?
Preferred Sources originated as, and remains most closely associated with, news and publisher use cases inside Top Stories and Discover. Its extension into AI Overviews and AI Mode broadens its relevance meaningfully, since these two surfaces answer a far wider range of query types than pure news queries — including the B2B research, comparison, and how-to queries covered throughout this series' vertical-specific guides. A B2B SaaS company with a genuinely engaged blog readership, an active newsletter, or a loyal returning audience segment has the same eligibility and the same underlying mechanism available as a news publisher, even though the feature's messaging and early adoption have skewed heavily toward traditional media.
The practical action for a non-news content operation: identify your highest-engagement existing audience touchpoint — likely an email newsletter or a returning-visitor segment identifiable in your analytics — and test placing the Preferred Sources deeplink there over the next quarter. Given the acknowledged low current adoption rate industry-wide, described across multiple sources as still in the low single digits of percentage terms for US users overall, an early, deliberate campaign targeting your most loyal existing audience faces meaningfully less competition for attention than the same ask will face once the tactic becomes standard practice across every content operation.
How NotioncCue Helps You Observe Whether Preferred Source Status Is Moving Your Citation Rate
Because Preferred Source status is inherently personalized — the badge and, per Google's confirmation, the citation-frequency benefit apply specifically to the user who selected you, not universally across all searchers — you cannot observe its effect through a single, generic prompt test the way you can test most other AEO variables covered in this series. What you can observe, over time, is whether your overall citation rate across your tracked prompt set trends upward as your preferred-source opt-in campaign gains adoption among your actual audience.
The NotioncCue Citation Tracker monitors your citation rate and the specific text of how you are described across ChatGPT, Perplexity, Claude, Gemini, and Google's AI surfaces on a consistent weekly cadence. Running a Preferred Sources opt-in campaign alongside this existing tracking gives you a before-and-after baseline: if your Google AI Overview and AI Mode citation rate on your tracked prompts shows a sustained upward trend correlating with your campaign's rollout and adoption curve, that is meaningful evidence the mechanism is working for your specific audience, distinct from whatever content or technical changes you may also be making concurrently.
Start your free NotioncCue trial and establish your current Google AI Overview and AI Mode citation baseline before launching a Preferred Sources campaign, so you have a genuine before-and-after comparison rather than an assumption about whether the tactic is contributing to any citation rate change you observe afterward.
Google has separately confirmed it is developing further extensions of the Preferred Sources concept, including exploring whether preferred status could eventually function as a more direct, universal ranking input rather than a personalized one. Treat the current mechanism as an evolving feature rather than a finished, stable specification, and revisit Google's own Search Central documentation periodically for updates to eligibility criteria and citation-weighting behaviour as the rollout continues through the remainder of 2026.
Frequently Asked Questions About Preferred Sources in AI Overviews and AI Mode
Can a publisher see how many users have marked them as a preferred source?
Not directly through any publisher-facing dashboard as of mid-2026. Google's own published figures (345,000+ total sources selected across all publishers, industry-wide) are aggregate statistics, not a per-publisher count made visible to individual site owners. The most practical proxy available is observing referral traffic and citation rate trends over time following a deliberate opt-in campaign, since no direct adoption-count reporting tool currently exists.
Does adding the "Add as preferred source" button to my site cost anything or require technical integration?
No. The button graphics and deeplink format Google provides are simple assets — an image and a URL — requiring no technical integration, API access, or approval process. Any site can implement this within an afternoon by adding the button and deeplink to existing high-traffic locations like article templates or email footers.
Is the Highly Cited badge something a publisher can request or apply for?
No. Unlike Preferred Sources, the Highly Cited label is calculated automatically by Google based on observed cross-publication citation and linking patterns, with no publisher-facing application or nomination process. The only lever a publisher has is producing genuinely original reporting or research substantive enough that other publications independently choose to reference and link back to it, which is the same underlying non-commodity content principle covered throughout the rest of this series.