For two years, links inside AI answers have been the black box of SEO. They appeared somewhere in the middle of the answer, in no pattern anyone could predict, and publishers watched their traffic thin out while Google said nothing measurable. Then on October 15, Google finally moved the levers: inline links in AI Mode and AI Overviews are going higher in answers and appearing more often.
This is the biggest link-visibility change Google has made to its AI surfaces — and unlike most AI-search news, it comes with a genuine action plan. Here is exactly what Google announced, what it changes for publishers, and five things you should do about it this month.
Key Takeaways
- On October 15, 2026, Google’s Hema Budaraju (VP, Product Management, Search) announced five link-visibility changes for AI Mode and AI Overviews: more inline links shown higher, a test of an “official sites” top section, Preferred Sources news carousels expanding globally, a redesigned “Dive Deeper” end section, and clickable creator/publisher profiles in testing.
- The changes matter because link placement inside an AI answer is the new position zero — higher, earlier links have the only real chance of converting a citation into a visit.
- Google shared no magnitude figures, no country list, no timelines, and no click data — two of the five items are tests, not launches. Treat this as direction-of-travel, not a traffic promise.
- The action items: build link-worthy, extractable content, make your official pages unmistakable, set up Preferred Sources presence, and measure with prompt portfolios — not assumptions.
What Exactly Did Google Announce on October 15?
The announcement came on Google’s official blog from Hema Budaraju, VP of Product Management for Search. Her framing: make it easy to find relevant sites, first-hand perspectives, and sources users love — “simpler, more visual, and more engaging.” Five concrete changes:
1. More inline links, shown higher. Inline links — links embedded in words or phrases inside the AI answer — have seen what Google calls “continued and substantial increases” in presence, and Google is launching updates to show them higher up in responses more frequently. Google gave no number for the increase. This is the core change: the citation is moving from buried to visible.
2. A test of an “official sites” top section. For task queries where only one or two official sites can complete the task or hold the exact info, Google will test a dedicated section at the top of the response listing those links, each with context on why it is useful. Google’s example: looking up how to get an airline refund. No launch date — it is explicitly a test.
3. Preferred Sources news carousels expand globally. The timely-news carousels inside AI answers are rolling out to more countries and languages (no list given) and will prioritize the user’s Preferred Sources. Google says Preferred Sources usage has grown more than 7x since the start of the year — it did not share the base number. This is the one change aimed squarely at publishers.
4. A redesigned “Dive Deeper” section. The related-article links at the end of AI answers get image thumbnails and a more clickable design. Announced, with no firm date — positioned as coming, not live.
5. Clickable creator and publisher profiles (testing). On AI Mode video results, creator and publisher names become clickable, linking to their Search profiles. In testing, with profiles opening in select countries.
Alongside, Google noted its News AI pilot licensing program — where it pays for content delivery and use to develop AI features — now covers more than 400 publications worldwide.
Why Does Link Placement Inside AI Answers Matter So Much?
Because a citation without a click is brand work, and a citation with a click is traffic — and placement decides which one you get.
Until now, links in AI Overviews tended to sit mid-answer or lower, surrounded by generated text that already answered the question. Most users never scrolled past the answer, so most citations earned zero visits. That is why the measurement discipline matters: tracking AI Overview citations taught us to separate citation presence from click-through.
Moving links higher and earlier changes the math. An inline link in the first paragraph of an answer is the only citation with a real shot at earning the visit — it appears before the answer has satisfied the query. The official-sites test section is even stronger: top of the response, with Google’s own context line explaining why your link is the right one.
But keep the scale honest. Google has not shared any click data for these changes, and independent reporting (Chartbeat data via trade press) shows publishers’ Google-referred traffic fell roughly 40% between July 2025 and July 2026. Higher placement is a visibility improvement to measure, not a traffic recovery to assume. The publishers who benefit will be the ones already structured to be cited — which is where the action items come in.
What Should You Do to Become Link-Worthy in AI Answers?
The announcement does not change what gets cited — it changes what a citation is worth. So the work is the same extractability discipline, now with higher stakes:
Build for inline extraction. Inline links attach to words and phrases inside the answer. Pages with quotable definitions, crisp one-sentence facts, and named entities give the system something to link to. A page that only has long prose gives it nothing to anchor on. Every key claim on your page should be expressible as a single linkable sentence.
Structure with tables and comparisons. The “X vs Y” and comparison tables that win featured snippets are the same structures that surface as linked passages in AI answers. If your content compares options in prose, rebuild it as a table and keep the prose below for depth.
Make definitions first, nuance second. Answers begin with “what is X” more often than not. A quotable definition in your opening paragraph is the easiest unit for both retrieval and inline linking. Writers build up to definitions; link-worthy pages lead with them.
Keep pages fresh and dated. In search mode, current sources win time-sensitive questions. Visible publication and update dates signal a maintained page — and a maintained page beats an identical abandoned one when the system picks what to link.
Match the rewritten query. When AI systems search, they typically rewrite the user’s question into targeted queries before retrieving. Your content must match the rewritten query, not just the conversational prompt — which means covering the question’s facets explicitly, not implying them.
How Do You Become the “Official Site” the Test Wants?
The official-sites test is the most interesting item in the announcement, because it has a clear winner profile: for task queries, Google wants the one or two official places that complete the task or hold the exact information.
That means making your official pages unmistakable:
One canonical page per task. If your site has three pages that could answer “how to reset X,” consolidate. The test rewards the single authoritative destination, not a cluster of near-duplicates.
Say what the page does in its first line. A page titled “Account Support” that opens with “Manage your account settings, billing, and security from one place” tells both the system and the user what task it completes. Google’s test section includes context on why each link is useful — pages that state their own purpose make that context easy to generate.
Kill ambiguity about who you are. Clear brand naming, consistent titles, and no competing unofficial pages. If affiliates, resellers, or aggregator pages rank for your task queries, your official page should be unmistakably official.
This is branded and transactional SEO, first. The test targets queries where the user wants the official source — logins, forms, downloads, policies. If your business lives on those queries, the official-site box is the highest-leverage placement in the entire announcement.
Should You Care About Preferred Sources?
If you publish news or timely content: yes, and probably more than you think.
Preferred Sources is Google’s feature letting users pick the outlets they want to hear from. With the news carousels expanding globally and prioritizing those picks — and usage reportedly up more than 7x since January — the carousel is becoming a curated front page inside AI answers.
The practical moves for publishers:
Make your outlet eligible. Preferred Sources works with recognized publishers. If your publication is not indexed as a news source, that is step zero — the standard news-publisher requirements apply before any AI-specific tactic.
Optimize for carousel eligibility. Timely articles with clear headlines, accurate timestamps, and proper article markup are the units the carousel pulls. Evergreen content will not appear here; this is a freshness surface.
Do not confuse the carousel with the inline links. They are separate placements with separate mechanics. The carousel rewards publisher authority and freshness; inline links reward extractable content. Most publishers need both.
If you are not a news publisher, Preferred Sources is someone else’s game — focus your energy on the inline-link and official-site work instead.
How Do You Measure Whether This Sends You Traffic?
The honest answer starts with what you cannot see. Google has shared no click data for these changes, and Search Console still has no AI-answer click dimension. You will not get a report that says “the October update sent you X visits.” So measure it yourself:
Run a prompt portfolio. Pick 20–30 questions your buyers ask, run them monthly in a clean session, and log: did an AI answer appear, were you cited, and where in the answer did your link sit — inline high, inline low, official section, carousel, Dive Deeper. Placement is now a metric, not just presence.
Watch AI referral traffic in GA4. Segment sessions from AI referrers (chatgpt.com, perplexity.ai, and direct Google AI surfaces where identifiable) using the GA4 AI-traffic method. This undercounts — many citations earn zero clicks — so treat it as a floor.
Track placement trends, not snapshots. The question is not “are we linked today” but “are our links moving higher in answers over the quarter.” Four-week trends are the signal; single checks are noise.
Build the dashboard. When the spreadsheet gets painful, the AI visibility dashboard method turns prompt portfolios into a client-ready reporting line: share of voice, placement distribution, and referral trend on one page.
The meta-lesson: Google’s announcement is a visibility change, and visibility is only valuable when measured. The sites that track placement will know within a quarter whether the October update moved their needle. The sites that do not will be guessing at the next announcement.
Final Thoughts
Google moving links higher in AI answers is the first real sign that the company wants AI search to send traffic back to the web — and the honest reading is that it is a beginning, not a reversal. Two of the five items are tests, none come with data, and the 40% referral decline did not happen overnight. But the direction is clear, and the playbook is clear with it: be extractable, be official where it counts, be present in Preferred Sources if you publish news, and measure placement like it matters — because for the first time, it might.
People Also Ask
Will the October 2026 update bring my traffic back?
Unknown — and Google has shared no data either way. Higher link placement improves the odds that a citation converts to a visit, but treat it as a visibility change to measure, not a traffic recovery to assume. Track your own placement trends and AI referral traffic.
Is the official-sites section live?
No — Google described it as a test, with no launch date. Build for it now (unambiguous official pages for task queries) so you are positioned if it launches.
What is the difference between inline links and the Dive Deeper section?
Inline links are embedded in the answer text itself, appearing where the claim is made. Dive Deeper is a separate related-links section at the end of the answer with thumbnails. Inline links are the higher-value placement because they appear before the user has finished reading.
Do I need to do anything new technically to be eligible?
No new files, markup, or machine-readable formats are required — Google has consistently said standard indexing and eligibility for Search snippets is enough. The work is content structure, not technical plumbing.
How is this different from AI Mode’s query fan-out?
Query fan-out is how AI Mode retrieves (breaking one question into many parallel sub-queries); this announcement is about how the resulting answers display links. The two connect: fan-out decides what gets considered, link placement decides what gets clicked. My guide to AI Mode keyword targeting covers the retrieval side.
