Google AI Mode Inline Links: What Publishers Should Change

Google is changing where links appear inside its generative search experiences. On October 15, 2026, the company announced more prominent inline links in AI Mode and AI Overviews, a test that surfaces official websites in a dedicated section, broader availability of timely-article carousels, more prominent related-article links, and clickable creator or publisher names on video results.

For publishers, the practical opportunity is not a new “GEO trick.” It is a wider set of source surfaces that still depend on discoverable, indexable and useful web pages. This guide turns Google’s announcement into an audit you can run without pretending that any tactic guarantees a citation.

What Google announced on October 15, 2026

Google’s announcement describes five changes across AI Mode and AI Overviews:

• More inline links: links embedded in words and phrases are appearing more often, and Google says it is showing them higher in responses more frequently.
• A dedicated official-site section: Google is testing a display near the top of some AI responses that highlights one or two official places to complete a task or get exact information, with context explaining why each link is useful.
• More timely-article carousels: prominent carousels are rolling out to more locations and languages globally and can highlight a user’s Preferred Sources.
• Richer related-article links: links at the end of some AI responses are getting image thumbnails and a more prominent design.
• Clickable creator and publisher names: Google started testing names on video results that open Search profiles in AI Mode.

Google also reported that the number of people using Preferred Sources had grown by more than seven times since the beginning of 2026. Treat that as a Google-reported adoption figure, not as proof that publishers will receive seven times more traffic.

What is fact: the product surfaces above and their rollout status come from Google’s announcement. What is inference: the prioritization and workflow below are editorial recommendations based on the eligibility rules Google publishes for AI features.

Why this update matters to publishers

The old mental model was “rank a blue link.” A more useful current model is “make the right page eligible and understandable when Google assembles an answer.” AI Mode and AI Overviews can use query fan-out, which means Google may issue several related searches across subtopics and data sources before composing a response. One user question can therefore create several opportunities for distinct pages to support different parts of an answer.

The October update increases the number and prominence of paths from that answer back to the web. It does not remove classic SEO constraints. Google’s documentation says a page must be indexed and eligible to appear in Search with a snippet before it can be shown as a supporting link. It also says there are no extra technical requirements or special schema required for AI Mode or AI Overviews.

This distinction is important. A page can be technically eligible and still never be selected. Google does not guarantee crawling, indexing or serving. The goal is therefore to remove preventable blockers and publish the kind of source that is genuinely useful for a particular subquestion.

Map each new surface to a real search intent

This mapping prevents a common mistake: rewriting every page into a generic question-and-answer template. The best asset for an official-site task is often a concise product or support page. The best asset for a timely carousel is usually reporting with a real publication date. The best asset for an inline citation may be one paragraph containing an original measurement, definition or procedure.

Run the eligibility audit first

Before changing copy, verify the basics Google explicitly names:

1. Indexing: use URL Inspection in Search Console to confirm the canonical URL can be indexed and to see the HTML Googlebot received.
2. Snippet eligibility: check that the page is not blocked by nosnippet, an overly restrictive max-snippet, or page-level controls that remove the useful passage.
3. Crawl access: confirm that robots.txt, your CDN and your hosting security rules allow Googlebot to fetch the page and its essential resources.
4. Internal discovery: link to the page from relevant, crawlable pages using anchors that describe the destination.
5. Text availability: make important facts available in HTML text, even when you also explain them in an image, chart or video.
6. Structured-data consistency: any structured data should match the visible page. It does not create a special AI eligibility shortcut.

Rubic8’s SEO Checker can help identify page-level basics such as titles, descriptions and headings. Its Robots.txt Generator and Tester can help you review crawler directives. These are diagnostic aids; use Search Console for Google’s authoritative indexing view.

Design passages that deserve an inline link

Google has not published a formula for winning an inline citation. The following is a reasonable content-design inference from how supporting links work:

• Give each section one clear job, expressed by an accurate heading.
• Answer the subquestion early, then supply the method, evidence and limitations.
• Prefer primary data, first-hand experience, original examples and exact definitions over a remix of existing summaries.
• Name the date and version when a fact can change.
• Link to the primary source so readers can verify the statement.
• Separate observed facts from your interpretation.

For example, “The feature improves discoverability” is vague. A stronger passage says what changed, where Google says it is available, which pages are eligible, and what remains unproven. That structure helps a person evaluate the claim even if no AI system ever cites it.

Avoid manufacturing dozens of thin pages for imagined fan-out queries. Query fan-out creates broader retrieval opportunities, but publishing overlapping pages can split internal signals and create cannibalization. Build a single strong guide when the intents share the same answer; create a separate page only when the task, audience or required evidence is materially different.

Strengthen official-page clarity without claiming “official” markup

Google describes its official-site section as a test for situations where one or two authoritative destinations can complete a specific task or provide exact information. It does not document a special tag that makes a page official.

A practical audit is still possible. Make the organization and page purpose unambiguous. Keep canonical product names consistent. Provide a direct route to core pages such as documentation, downloads, pricing, account access and support. Consolidate duplicate versions. Use one canonical URL for the definitive answer, and make ownership and contact information easy to find.

Those steps are good information architecture, not a guarantee of placement. Do not add “official” repeatedly to titles or headings merely to target the display. If the page is a reseller, aggregator or independent review, describe that relationship honestly.

Treat freshness as evidence, not decoration

Prominent article carousels create an obvious opportunity for newsrooms and specialist publishers, but changing a timestamp without changing the page does not make the reporting fresh. Use a visible “updated” date when you made a substantive revision, and explain the revision when it materially changes the story.

For fast-moving topics, add the event date in the introduction, verify time zones, preserve the original publication date, and state which specifications were current when you checked them. If a source corrects or retracts a claim, update both the claim and the context around it. Accurate chronology is especially valuable when an AI response combines pages from different moments.

Use Preferred Sources as an audience feature

Preferred Sources lets a reader select a publication. Google says that selected sites are more likely to appear in Top Stories with a preferred badge, and can be highlighted with that badge in AI Mode and AI Overviews where those features are available.

Eligibility is domain- or subdomain-level, not subdirectory-level. For AI features, Google says the site must be included in Search generative AI features in Search Console. Publishers that appear in the source-preferences tool can use Google’s standard button, a custom implementation or a deeplink. Google explicitly says these promotional methods are optional and are not required to appear as a preferred source.

Do not hide the control behind a misleading label or imply that selecting the site guarantees rankings. Place it where returning readers already express loyalty: near a newsletter sign-up, on an author page, or after a useful article. Measure completed preference flows and returning-reader behavior, not just button impressions.

Audit publisher and creator identity separately

The creator-profile test is most relevant to video publishers. A Google Search profile can aggregate work from a website and social platforms, but availability and eligibility requirements vary. Google’s October announcement points eligible creators and publishers to the Search profile setup flow.

Check whether an eligible profile already exists before trying to create one. Keep names, handles, bios and linked properties consistent enough for people to recognize the same entity. If you add a Search profile badge, use Google’s published badge assets and brand guidelines. A Search profile is a distribution and identity surface; Google’s help documentation says creating one does not directly change Search rankings.

Build images for the related-article surface

Google says some related-article links will use thumbnails and a more clickable design. That makes image quality an editorial requirement, but not a reason to stuff text into every thumbnail.

• Use an image that conveys the subject at small sizes.
• Provide descriptive alt text for accessibility.
• Serve responsive dimensions and avoid unnecessary bytes.
• Keep the image crawlable and associated with the visible article.
• Avoid generic stock art when an original diagram, screenshot or data visual would be more useful.

The same asset should still work on a normal search result, social share and article page. Design for the reader first rather than reverse-engineering one experimental placement.

Measure visibility without inventing an AI citation KPI

Google’s June 3, 2026 announcement introduced dedicated Search Generative AI performance reports in Search Console, and an update note says the insights reached all websites worldwide on August 31. The dedicated views cover impressions, pages, countries, devices for Search, and trends over time. Google also states that this data remains included in the overall performance report.

Use the dedicated report to identify which URLs receive impressions in generative AI features, then join that observation with ordinary Search Console and analytics data. A sensible review includes:

• Generative AI impressions by page and country.
• Landing-page clicks and changes in the overall Web report.
• Engaged sessions, newsletter sign-ups, trials or other meaningful conversions.
• Visibility changes after a real content or technical improvement.
• Manual spot checks recorded as qualitative examples, not as a benchmark.

Do not report “AI citation share” from a handful of prompts as if it were stable market share. Responses vary by query, model, location, time and user context. For a broader measurement framework, see Rubic8’s guide to the Google Search Console multimodal report.

A 30-minute publisher workflow

1. Choose one important query family. Write down the main task and three genuine subquestions.
2. Assign one canonical page to each distinct intent. Merge or redirect pages that answer the same task.
3. Inspect eligibility. Check indexing, canonical selection, snippet controls, robots.txt and internal links.
4. Improve the source passage. Add the direct answer, original evidence, date, limitations and primary-source link.
5. Review identity and freshness. Make author, publisher, publication date and meaningful updates visible.
6. Check the image. Confirm relevance, crawlability, alt text, dimensions and file weight.
7. Open the generative AI report. Record the current pages and impressions so you have a baseline.
8. Revisit after recrawl. Evaluate both visibility and reader outcomes; do not judge the change from one prompt test.

What not to do

• Do not create an “AI version” of every existing article.
• Do not add unsupported schema or an AI-specific text file and call it a requirement.
• Do not block Googlebot while expecting inclusion in Google Search AI features.
• Do not confuse Google-Extended, which concerns training and grounding in certain other Google systems, with Googlebot controls for Search.
• Do not present a vendor-reported adoption metric as a publisher traffic forecast.
• Do not promise inclusion in the official-site section, carousels, inline links or profiles.

Google’s own 2026 generative AI optimization guidance emphasizes valuable, unique and non-commodity content while mythbusting common AEO and GEO claims. Rubic8’s Google AI content guidelines article explains how that principle applies to AI-assisted publishing workflows.

The practical takeaway

Google’s October 15 update gives publishers more visible routes from an AI answer to the open web. The opportunity is real, but it is not a separate optimization system. Pages still need to be crawlable, indexed, snippet-eligible, internally discoverable and genuinely useful.

Start with the surface that matches your audience: inline evidence for evergreen explainers, canonical destinations for official tasks, accurate chronology for news, strong images for related articles, and audience controls for Preferred Sources. Then measure visibility in Search Console and business value in analytics. That is less exciting than a secret GEO hack—and much more defensible.

Official sources

• Google: More ways to connect with the web from Search — October 15, 2026.
• Google Search Central: AI features and your website.
• Google Search Central: Preferred Sources for web publishers.
• Google Search Central: Search Generative AI performance reports — June 3, 2026; global rollout note dated August 31, 2026.
• Google Search Central: A new resource for optimizing for generative AI in Search — May 15, 2026.
• Google Search Central: Add a Search profile badge.