Google's New 'Preferred Sources' Button Lets Publishers Fight Back Against AI Traffic Losses

For years, the bargain between Google and the open web was simple: publishers let the search engine index their content, and Google sent them traffic in return. AI-powered search has broken that bargain. With AI Overviews and AI Mode answering questions directly on the results page, clicks to publisher sites have collapsed — with some outlets reporting referral declines of 20% to as much as 90%. This week, Google made its most direct concession yet to the industry it disrupted: an embeddable “Preferred Sources” button that turns readers themselves into a distribution channel for the sites they trust.

What Google announced

On August 20–21, 2026, Google introduced a new interactive “Preferred Sources” button that online publishers can embed directly on their own websites. When a reader clicks it, the site is immediately added as a Preferred Source on Google, and the reader is bounced right back to the exact spot on the page where they left off — no interruptive detour, no settings maze.

Once a site is marked as preferred, its content is more likely to surface across virtually every surface where Google distributes content: Top Stories in Search, AI Overviews, AI Mode, Discover, and Google News. Google’s pitch is that this makes it “easier for readers to find links from the sites they know and trust,” whether they’re running a classic search query or conversing with Google’s AI about a topic.

The feature builds on groundwork laid in May 2026, when Google first brought Preferred Sources to its AI experiences after an earlier debut in Top Stories. At that point, Google said users had already selected over 345,000 unique sources. In this week’s announcement, that number has climbed past 600,000 — evidence, Google argues, that readers actively want a say in which publications reach them.

The button’s embed code is available now in Google Search Central documentation, and readers who never see a button can still add favorites manually via Google’s source preferences page by searching for a publisher by name or website.

Why it matters: the click-through numbers

The commercial logic behind the button is straightforward. Google’s own earlier studies found that people are twice as likely to click through to a preferred source when one is available in results. Doubling click-through rates on a subset of results won’t restore the traffic publishers have lost, but it gives newsrooms a concrete mechanism to convert loyal readers into a durable ranking signal — something that has historically been impossible to buy or engineer.

It also arrives amid an increasingly hostile environment for traffic-dependent businesses. Axios noted in July that “Google has basically stopped sending people to websites for answers and information.” A May 2026 analysis found AI Overviews had cut publisher click-through rates by 58%, prompting Google to add five traffic-sending features to the format. The embeddable button is the latest — and most publisher-friendly — of these gestures, though critics note it still leaves the fundamental economics of AI summarization untouched: Google keeps the user on Google.

Beyond the button: natural-language feed tuning

The Preferred Sources button shipped alongside two more personalization features that reveal where Google thinks search is heading.

First, in the coming days, the Google app’s Discover feed will let users tune their feed in their own words. Tapping any three-dot menu on a card opens a prompt where readers can type what they want more or less of — “kitchen renovation ideas, eco-friendly only” or “road trip videos, but no camping” — and the feed adjusts in real time, remembering the request going forward. Google describes the goal as making feed tuning “as easy as texting a friend.” The move mirrors the wave of user-controlled algorithms that major social platforms have shipped in recent months, as recommendation systems across the industry shift from opaque personalization to explicit, conversational control.

Second, on Android, the Google News app now supports customizable audio daily briefings. Users can curate briefings by choosing specific topics, with clear source attribution and links out to full articles — a small but notable nod toward sending listeners back to publishers rather than keeping them inside a closed audio experience. Google also says that thanks to the publishers participating in its news AI pilot program, briefings will increasingly include “deeper dives” into key stories.

Analysis: a lifeline, not a rescue

The strategic read here is twofold. For Google, Preferred Sources generates a rich, explicit signal about source trustworthiness — data that almost certainly also improves the citation behavior of AI Overviews and AI Mode, where Google has been under pressure to attribute and link out more generously. Every reader who clicks a publisher’s button is, in effect, casting a vote Google can fold into its ranking and citation systems.

For publishers, the button is best understood as a new channel with real but bounded upside. Being someone’s preferred source doubles the odds of winning a click, and 600,000 sources have already been chosen organically. But the feature does nothing to slow the broader migration from ten blue links to AI-generated answers, and it implicitly accepts that the default experience — for users who never touch a preference — remains one where Google’s AI, not the publisher’s page, is the destination.

There’s also an uncomfortable power dynamic embedded in the design: the only way for a publisher to participate is to send readers to Google to register a preference. Publishers must now actively funnel their own audiences into Google’s settings — a striking inversion of the old traffic bargain.

Still, for an industry that has spent two years watching referral graphs fall off a cliff, an official Google feature whose stated purpose is to drive more clicks to publisher websites is a genuine change in posture. Publishers who move early — embedding the button prominently, explaining it to readers, and pairing it with the new Discover and audio-briefing surfaces — will have a first-mover advantage in rebuilding the direct reader relationships that AI search has been quietly dissolving.