Google is rolling three personalization features across Search, Discover and Google News, and the most consequential one is not aimed at readers at all. It is a publisher-facing button designed to pull audience relationships back inside Google’s own surfaces, right as those surfaces increasingly answer queries without sending a click anywhere.
Per Google’s own announcement, written by Mrinalini Loew, General Manager of Google’s Search Ecosystem, the company is shipping an embeddable “Preferred Sources” button publishers can place on their own pages. Clicking it adds that site as a preferred source and returns the reader to where they left off, and Google says preferred sources are “more likely to appear in Top Stories” and can carry a preferred badge inside AI Overviews. Google states that “people have already selected more than 600,000 unique sources,” up from 345,000 in May. Separately, Google app users will soon be able to type natural-language requests to reshape their Discover feed, and Google News on Android now lets listeners customize daily audio briefings by topic, with source attribution and links to the full articles built in.
Why it matters: this is Google building a first-party preference layer for an era where its own AI surfaces are the thing standing between a publisher and a reader. A preferred-source signal that publishers themselves have to solicit, one click at a time, is a meaningfully different relationship than an algorithmic ranking signal, and it gives Google a defensible answer to the argument, already made in this publication, that AI answers are erasing brand visibility that used to come from a ranked link.
The original insight: 600,000 sources sounds broad until set against how few brands are positioned to use the button well. The mechanism rewards sites with existing loyal-reader habits, since a reader only clicks “prefer this source” for something they already return to, which means Preferred Sources is likely to concentrate visibility further toward publishers who were already winning distribution, the same dynamic recent research on AI assistants and local business visibility found working against smaller, less-established sites.
Source: Google
