Google's Highly Cited Label: How to Earn One
The Preferred Sources rollout into AI Mode and AI Overviews got most of the coverage on May 27, 2026. The same announcement introduced a quieter feature that's arguably more important for SEO long-term: the "Highly Cited" label. Google's product manager Duncan Osborn confirmed the label appears on article links that other publishers reference most, calculated from cross-publication linking patterns.
Brass-SEO's view is that Highly Cited is the more durable signal of the two May 2026 features. Preferred Sources is user-driven and resets when a user clears preferences. Highly Cited is publication-driven, accumulates with original reporting that genuinely earns citations, and serves as an authority signal that travels with the article across AI surfaces. This post covers what it is, how Google calculates it, what earns one, and the limits of pursuing it as a tactic.
This is the companion piece to Google Preferred Sources: How to Add the Button. Both features landed in the same May 27, 2026 announcement.
Quick Navigation
- What "Highly Cited" Means
- How Google Calculates It
- Where the Label Appears
- How Highly Cited Differs From Preferred Sources
- What Earns a Highly Cited Label
- A Realistic Publisher Strategy
- The Limits of This Tactic
- Frequently Asked Questions
What "Highly Cited" Means
The Highly Cited label is a visible marker Google attaches to article links that other publishers and reporters frequently reference. Google's May 27, 2026 announcement from Duncan Osborn, Product Manager at Google Search, framed the label as helping readers identify "the story other stories are citing" — primary reporting that other coverage builds on.
Brass-SEO reads the underlying mechanism as a cross-publication citation graph. When a story breaks and dozens of other publications link to a single article as the source, Google's systems recognize that article as the primary reference and apply the label. The exact threshold isn't published, but the directional signal is clear: scarce, original primary reporting earns the label; secondary coverage that aggregates others doesn't.
The label isn't a ranking signal in the traditional sense. It's a visibility marker inside AI Overviews, AI Mode responses, and Top Stories carousels that tells readers the source is widely referenced. The effect on click-through is similar to the Preferred Sources badge, but the path to earning it is harder.
How Google Calculates It
Google has not published the exact algorithm for Highly Cited determination. The pattern from the May 2026 announcement and follow-up coverage points to cross-publication linking as the primary input. Brass-SEO's interpretation, based on the published statements:
- Inbound citation density. Pages with many independent publishers linking to them as the source for a claim earn higher Highly Cited scores
- Temporal clustering. Citations earned in a short window after publication (the news cycle for a story) appear to weight more heavily than diffuse citations spread over months
- Topic specificity. The label appears more often on stories with clear topic boundaries (investigation findings, original research, exclusive interviews) than on general explainers
- Publisher diversity. Citations from many distinct publishers seem to count more than many citations from one or two referrers
The mechanism rewards original reporting at the moment of publication, not evergreen authority built up over years. This makes the label different from traditional E-E-A-T signals or domain authority — it's story-level, not site-level.
For the broader context on how citation patterns now drive AI search visibility, see Why ChatGPT Cites Some Pages and Skips Others.
Where the Label Appears
Highly Cited labels appear in three Google surfaces as of May 2026:
- AI Mode responses. When AI Mode cites multiple sources for an answer, the Highly Cited source gets a visible badge
- AI Overviews. Same mechanism inside the AI Overview summary at the top of traditional SERPs
- Top Stories carousels. Visible on the standard news carousel that appears on many query types
- "Perspectives" carousels. A new article carousel format for developing topics, where Google surfaces original reporting alongside Preferred Sources
Google also confirmed the announcement that the label travels with citations. When a story referenced as Highly Cited gets quoted in a second article, Google will indicate that the second article references a Highly Cited source. This is the closest the system gets to a transitive authority signal.
How Highly Cited Differs From Preferred Sources
Both labels landed in the same May 2026 announcement but they work through opposite mechanisms. The distinction matters for publisher strategy.
| Trait | Preferred Sources | Highly Cited |
|---|---|---|
| Who controls it | The user (via personalization) | Google (via citation graph) |
| Publisher action to earn | Optional button drives adoption | No direct action; earned through original reporting |
| Reset condition | User clears preferences | Article ages out of news cycle |
| Per-article or per-site | Per-domain | Per-article |
| Time to earn | Minutes (one user click) | Days to weeks (cross-publication citations to accumulate) |
| Strategy implication | Audience development | Original reporting investment |
The two labels can co-occur on the same citation if a user has selected the publisher AND Google considers the specific article Highly Cited. Brass-SEO has observed this stacking in AI Mode responses during May 2026 testing.
The strategic split: Preferred Sources rewards owning an audience and inviting them to designate you. Highly Cited rewards being the source other publishers reference. Both are valuable; they require different investments.
What Earns a Highly Cited Label
Brass-SEO's reading of the announcement and the patterns observed in AI Mode responses points to four categories of content that consistently earn Highly Cited labels:
- Original investigations. Multi-week reporting projects with primary sources. These get cited by every publisher that covers the story.
- Exclusive interviews and quotes. When you talk to the named source first, other publications cite you when they quote that source secondhand.
- Original research and data studies. Survey results, dataset analyses, internal analytics teardowns. Other publishers reference your numbers.
- First-hand accounts of newly-launched products or features. Being the first to publish a detailed analysis of a new product release earns citations from later coverage.
What does not earn Highly Cited labels, based on observation: roundups, summaries of existing coverage, opinion pieces, explainers of well-known concepts, and reformatted versions of others' reporting. These are useful content categories for traffic but they're consumers of citations, not producers.
For more on original data as a citation driver, see the relevant section in Why ChatGPT Cites Some Pages and Skips Others and Original Data SEO: Real Results from a Case Study.
A Realistic Publisher Strategy
Most small publishers won't earn Highly Cited labels at the scale of national news outlets, but the label is per-article, not per-publisher. A single piece of original reporting can earn the label even on a site that mostly publishes explainers. Brass-SEO recommends a portfolio approach.
The realistic strategy for a small to mid-sized publisher in 2026:
- Publish 80% explainer/evergreen content that drives steady traffic and supports your topical authority
- Reserve 20% for original-reporting projects — interviews, surveys, data analyses, first-hand product walkthroughs
- For each original piece, run targeted outreach to the publications most likely to cite it, with a short pitch that includes the quote or stat you want them to use
- Measure citations earned per piece using Google searches for distinctive quotes or phrases from your article (the "named expression" tracking pattern)
The portfolio approach lets you compete for Highly Cited labels on the 20% pieces while maintaining traffic from the 80%. A site that publishes 100 explainers per year and 25 original-reporting pieces typically earns more Highly Cited labels than a site that publishes 125 explainers, even though total volume is identical.
For the underlying outreach mechanics, see Earn Backlinks by Correcting Grokipedia Errors for one example tactic and Link Building Without Cold Emails for the broader framework.
The Limits of This Tactic
Brass-SEO sees four real limits to pursuing Highly Cited labels as a strategy:
It rewards a content type most publishers don't produce. Original reporting is expensive in time and craft. The cost-per-piece for a real investigation is 5-10x a competent explainer. Many sites are right to skip the category entirely and focus on tactics with better unit economics.
The label is ephemeral by design. Stories age out of the news cycle. A piece that earns the label in June may not carry it in September. Unlike traditional backlinks, which accumulate as a durable authority signal, Highly Cited is story-specific and time-bound.
Verification is hard without Search Console support. Google hasn't added a Highly Cited report to Search Console as of May 2026. Publishers can observe the label on their own articles in AI Mode and Top Stories, but tracking which pieces earned it and which didn't requires manual checking.
The competitive set is consolidating. Wire services, established news outlets, and a small number of specialized publications already dominate Highly Cited labels in many topic areas. Breaking into the cited set requires either picking under-covered niches or genuinely scooping established outlets — both are harder than the alternative tactics.
For most small businesses and solo publishers, the right call is to skip pursuing Highly Cited as a primary tactic and focus on Preferred Sources (audience-driven) plus citation-structure optimization on the content you already publish. See Why ChatGPT Cites Some Pages and Skips Others for the citation-structure side.
To audit your own pages for AI citation patterns and identify which ones are worth the original-reporting investment, start a Brass-SEO trial.
Frequently Asked Questions
Is Highly Cited the same as a backlink?
Not exactly. Backlinks are direct HTML links from one site to another. Highly Cited reads citations broadly, which includes traditional backlinks but also brand mentions, attribution to source quotes, and references to original data. Google's specific algorithm isn't published, but the directional signal is that the citation graph extends beyond just <a href> tags.
Can I earn the label without traditional press coverage?
Possible but harder. The vast majority of Highly Cited labels Brass-SEO has observed in AI Mode results go to articles that earned traditional press citations from established publishers. Niches where you're a recognized expert and where peer publications cite each other are the realistic targets for non-press publishers.
Does the Highly Cited label boost my Google rankings?
Google hasn't characterized it as a ranking signal in the traditional sense. The published framing is visibility in AI Mode, AI Overviews, and Top Stories carousels. Indirectly, citation density that earns the label also correlates with the kinds of signals Google's ranking systems reward (authoritative content, named expert references, original sources).
How do I know if my article earned the label?
Manual check as of May 2026. Search for queries where your article might be cited, look at the AI Overview and Top Stories results, and inspect the displayed labels. Brass-SEO recommends running this check on the 20% original-reporting pieces every 2-3 weeks during their first 90 days of life.
What's the relationship between Highly Cited and Brass-SEO's AI Citability audit?
Different layers. The Brass-SEO AI Citability audit checks structural traits (answer capsules, link-free capsules, named expert quotes) that make a page extractable by AI surfaces. Highly Cited rewards earning external citations once the page is published. The audit gets you into the citation-eligible set; the original-reporting strategy earns external references that produce the label.