Google's Preferred Sources puts the burden on the people who won't carry it

Google just shipped a feature called Preferred Sources. It lets you tell Google which websites you trust, and the AI Overviews sitting at the top of your search results will weight those sites more heavily. On paper it sounds like a reasonable concession to users who are tired of AI summaries built from low-quality content. In practice, it hands the problem back to the exact people who are least equipped to solve it.

The feature is opt-in. It lives inside Search settings. You have to know it exists, navigate to it, and then manually add domains. That is three steps most users will never take. Not because people are lazy, but because that is not how people use Google. They open a tab, type a question, and read what appears. The idea that a meaningful share of Google's user base will start curating a personal allowlist of trusted domains is a design assumption that has no basis in how the product is actually used.

The people who need this most will benefit the least

Think about who actually changes browser settings. Or who has a LastPass account. Or who has ever visited chrome://flags. It is a small, technically curious slice of the population. These are also the people who already know how to evaluate sources, who use browser extensions, who can spot a content farm. They are not the users being misled by AI Overviews pulling from thin, ad-stuffed aggregators.

The users who would genuinely benefit from a Preferred Sources list are people who do not know what a domain is in the sense that would allow them to curate one. They trust Google to surface the best result. That trust is exactly what AI Overviews have been quietly eroding, and Preferred Sources does nothing to address that at scale.

This is a pattern worth naming. When a platform ships a settings-based solution to a systemic quality problem, it is not solving the problem. It is documenting that the problem exists while moving responsibility off the platform and onto the user. Google has done this before with SafeSearch, with ad personalization controls, with News source preferences. The controls exist. Almost nobody touches them.

What this means for publishers and content-driven businesses

If you run a brand that depends on organic search, or you advise clients who do, the honest read of Preferred Sources is this: it will not restore the traffic that AI Overviews have compressed. A 2024 Semrush study found that AI Overviews appear in roughly 13 percent of all search results in the US, with higher rates in informational and educational queries. Those are exactly the queries where publishers have historically earned clicks. The overview answers the question. The user moves on.

Preferred Sources does not change that mechanic. Even if a user has added your domain to their list, the AI Overview still runs. Your content might inform the summary. Your site might get a citation link. But the click often does not happen, and a citation in an AI summary is not the same as a reader on your page.

For agencies advising clients on content strategy, this is the moment to stop optimizing for a version of Google that no longer exists. The question is not how to rank in position one. It is whether your content has a reason to exist that does not depend on Google deciding to show it.

The deeper issue with opt-in quality controls

There is a structural tension in how Google has approached AI Overviews from the start. The feature is on by default for most users. Opting out requires finding a Labs setting or using a filter. Preferred Sources is additive to a system you cannot easily turn off. That asymmetry is not accidental. Default-on features accumulate data, train models, and generate engagement metrics that Google needs to justify the investment. Opt-in controls for quality give the appearance of user agency without meaningfully shifting the experience for the 99 percent who never configure anything.

This is not a conspiracy. It is just how large consumer products work. The default state serves the platform's goals. The settings serve the optics. What is worth watching is whether regulators, particularly in the EU under the Digital Markets Act, start treating default-on AI summaries that suppress publisher traffic as a market conduct issue rather than a product design choice. That framing is already emerging in some early DMA enforcement conversations around Google's search presentation.

The distinction matters because a regulatory constraint changes the default. A settings menu does not.

What to actually do with this

For studios and agencies advising content or media clients, a few things follow from this.

First, stop treating GEO as a backup plan for SEO. Generative Engine Optimization, the practice of structuring content so AI systems cite it, is real and worth doing. But it optimizes for citation, not for visits. If your business model needs the visit, citation is a consolation prize.

Second, reconsider the role of owned channels. Email lists, community platforms, and direct relationships with readers are not nostalgic alternatives to search. They are the assets that do not depend on a third-party algorithm deciding what to surface this quarter.

Third, be direct with clients about what Preferred Sources actually is. It is not a fix. It is a pressure-release valve that Google has shipped so it can point to user controls when asked hard questions about search quality. Your clients' audiences will not use it. Their competitors' audiences will not use it. The competitive landscape around organic search does not shift because of this feature.

The work is the same as it was six months ago. build content that earns attention from the people who are actively looking for what you make, through channels you at least partially control. Google remains useful for discovery. It is just no longer reliable as a distribution strategy on its own, and Preferred Sources is not going to change that.

The short version

Google's Preferred Sources is a well-intentioned feature with a fatal assumption: that users will use it. Most will not. The systemic problem, AI Overviews trained on or weighted toward thin content, remains untouched for the vast majority of searches. Publishers lose traffic. Users get summaries of uncertain quality. And Google gets to point at a settings page and say the choice is yours.

If you are running or advising a content-driven business, treat this feature as a signal, not a solution. The signal is that Google knows AI Overviews have a quality problem. The solution is not going to come from a settings menu.