On 15 October Google announced five changes to the way AI Mode and AI Overviews link back to the web. The post, More ways to connect with the web from Search, is short on numbers and long on direction: more inline links, a section for official pages, wider news carousels, more prominent related-article links, and clickable creator names on videos.
Publishers have heard "more links" before and want the click data. Agencies started asking which of it they can act on for clients.
That second question has a concrete answer. Much of what Google described routes through surfaces with real eligibility requirements attached. Two of those are structural, checkable in about ten minutes per client, and a meaningful number of sites fail them without anyone noticing.
What actually shipped, and what is still a test
Google's own wording matters here, because the five items sit at different stages:
- More inline links. Launching. Google says it has made "continued and substantial increases" to inline link presence, plus updates to show them higher in responses more often. No volume figures given.
- A dedicated section for official pages. A test. Google says it "will be testing" a display that surfaces the one or two official places to complete a task, at the top of a response, with context on why each link is useful.
- Timely-article carousels. Expanding to more locations and languages, and highlighting a user's Preferred Sources.
- Related-article links. A design change: image thumbnails and a more clickable treatment at the end of some responses.
- Clickable creator and publisher names on video results. A test that started the week of the announcement, in AI Mode only, linking to Search profiles.
So: two launches, one expansion, one design change, two tests. An announcement does not mean every searcher sees every format, and Google published no click-through data alongside any of it. The one hard number is about user behaviour rather than traffic: Preferred Sources usage has grown more than 7x since the start of the year.
That number is the thread worth pulling, because Preferred Sources now runs through several of these surfaces.
Gate one: Preferred Sources works at the domain level only
When a reader selects your site as a preferred source, your content is more likely to appear in Top Stories with a "preferred" badge, and can carry that badge in AI Mode and AI Overviews too.
The eligibility rule is in Google's publisher guide, and it is specific: only domain-level and subdomain-level sites appear in the source preferences tool. Google's own example is that example.com and code.example.com are eligible, but the subdirectory example.com/blog is not.
For a client publishing on their own domain this is a non-issue. The domain is the unit of selection, so the blog comes along with it.
The problem case is the one agencies build all the time: a client's content hub in a subdirectory of somebody else's domain. A Medium publication, a marketplace profile, a parent-company site where the acquired brand got a folder rather than a subdomain. In all of those the opt-in accrues to the platform, not your client. You can run the campaign perfectly and the reader still subscribes to the host.
If a client's editorial sits in a folder on a domain they do not control, that is now a visibility decision rather than a tidiness one.
Gate two: the Search generative AI control
The second gate is a Search Console setting most people have never opened.
The Search generative AI control governs whether your site's links and content can appear in AI Overviews, AI Mode, and generative features in Discover. Google rolled it out to all websites worldwide on 31 August 2026. The default is to include your site, so this is not a switch anyone needs to flip to be visible.
The catch is inheritance. A property inherits its control from its closest parent that stopped inheriting. If anyone set exclusion at a domain or URL-prefix property above the one you look at, the child follows it silently. Nobody gets an alert on the child, and the only notification goes the other way: owners of the top-level domain see a notice when a child overrides them.
Google is also explicit that this control is required for the badge: to show as a preferred source in AI Mode and AI Overviews, your site has to be included in Search generative AI features.
Two more details worth holding onto. The control is not a ranking signal and does not affect the rest of Search. And it is separate from AI training, which is governed by Google-Extended, so a decision about training does not describe your AI Search visibility.
For an agency running thirty client properties, this is a half-hour audit, once.
The creator profiles test is narrower than it reads
The clickable creator names item got the most attention and applies to the fewest readers. Search profiles have published requirements: they are currently available in Canada, India and the United States only, the person must be at least 18, and they need 10,000 subscribers or followers on at least one platform (YouTube, Instagram, X or TikTok). Google also states plainly that creating a Search profile does not directly affect ranking.
For a B2B SaaS client or a regional services business, this is not a near-term lever. Treat it as a signal about where Google is taking source attribution, not a task for this quarter.
What to do with this
The honest version: Google restated a direction it has moved in all year, and the actionable part is plumbing, not strategy.
- Check whether each client domain appears in the source preferences tool. If not, the usual cause is thin publishing frequency, not a penalty.
- Find any client whose content lives in a subdirectory of a domain they do not own, and treat migration to their own domain or subdomain as a visibility item, not a backlog item.
- Audit the Search generative AI control across the portfolio, parent properties included, and check what each child inherits.
- Keep publishing on a consistent cadence. These surfaces still select from content that exists and is relevant to the query. None of them conjure a citation for a page you did not publish.
That last point is where the real work sits. Preferred Sources is a tiebreaker among sites that already publish relevant material, not a substitute for having any. The sites that benefit from more link slots will be those with something on their own domain worth linking to, published often enough that Google treats the domain as active.
One separate note: October's reported ranking volatility is not a confirmed update. As of 19 October the Search Status Dashboard still listed the September 2026 spam update, completed on 8 October, as the most recent ranking incident, according to Search Engine Watch. Movement is not evidence of an update, and the two stories are best kept apart.
FAQ
Does my site need special schema to appear in these new AI Mode link formats?
No. Google has not published an AI-Mode-specific schema specification, and there is no markup that flags a page for inclusion. Standard structured data helps machines resolve what your page is about and who published it, but there is no AI Overviews schema type to implement.
My client's blog sits in a /blog folder. Is that a problem for Preferred Sources?
No. Eligibility is assessed at the domain and subdomain level, so a blog in a folder on your client's own domain is covered by the domain entry. The problem is only when the content sits in a subdirectory of a domain your client does not control, because then the opt-in belongs to whoever owns that domain.
How do I check whether a site is excluded from AI Search features?
Open Search Console, go to Settings and then Search generative AI. The page shows whether the property is set to include, exclude, or inherit from its parent. Check the parent properties too, because a child inherits silently and its owner gets no notification.
Does setting that control to exclude hurt rankings?
Google states the control is not used as a ranking or inclusion signal affecting other parts of Search. Excluding removes your content from generative AI features and the traffic and impressions that come with them, but it is not described as a wider ranking consequence.
How quickly does a change to that control take effect?
Google says exclusion generally takes a few days, with content excluded within one to two days after the control goes live, though some content can take longer because of caching and propagation across Google's systems.
Is any of this worth prioritising over ordinary SEO work?
Only the audit parts, because they are cheap. Confirming eligibility and checking a Search Console setting across a portfolio takes an afternoon. Everything else rewards what it always has: content on a domain you control, published consistently, that answers the query better than the alternatives.