Definition: what is Preferred Sources?
A user marks a website as a preferred source — and for them, and only for them, serving changes: the source appears more likely in Top Stories and is labelled with a “Preferred” badge there and in AI Overviews and AI Mode. The 2026 development came in three documented steps: launch for Top Stories, extension of the effect to the generative surfaces, and since 20 August 2026 the option of a native button on your own website.
Only domains and subdomains are eligible — subdirectories cannot be marked separately. The feature is available for Top Stories globally, in all languages of Google Search.
How it works — and how it does not
The effect is strictly personalised: there is no global ranking boost; one user's choice changes nothing for others. Preferred Sources is personalisation, not an authority metric — a campaign for “more markings” does not lift general rankings. Google publishes no figures on effect size; how often a preferred source appears additionally and how the badge changes click behaviour is unproven. The feature is loyalty infrastructure with an open effect coefficient, not a calculable traffic channel.
Strategically the direction is remarkable: for the first time an explicit user decision influences which sources a generative answer carries and how they are labelled — a user-driven channel emerges beside purely algorithmic source selection. Brand loyalty now pays directly into generative visibility.
Embedding: deeplink or button
Deeplink (recommended start): a normal link to https://www.google.com/preferences/source?q=yourdomain.com — no script, no third-party request on page load, no new consent category. Data processing starts only with the user's active decision on Google's own surface.
Official button: Google provides a script from news.google.com that renders a native button. The price: data flows to Google on page load — in most consent setups this is a consent-requiring third-party embed. Sensible only where consent management is well maintained and the visual prominence justifies the effort. Functionally both paths lead to the same marking.
Who benefits
The feature rewards a specific relationship: returning readers who know a source by name. Core cases are publishers and news offerings (Top Stories as the native stage) plus trade blogs and corporate content with frequency and a regular readership. Without returning readers there is nothing to gain — the feature extends existing loyalty into Search; it does not create it.
Related terms
Preferred Sources acts in AI Overviews and AI Mode and is thus part of the generative visibility mechanics. The full analysis including the consent trade-off: Google Preferred Sources.
Loyalty becomes a serving signal
Preferred Sources translates reader loyalty into serving for the first time in documented form — personalised, without a global boost. If you have a readership, let them mark you; if you have none, there is nothing to gain.
FAQ on Preferred Sources
Does Preferred Sources improve Google rankings? ▾
No. The effect is strictly personalised: only the marking user sees the source more likely and with a badge — in Top Stories, AI Overviews and AI Mode. Nothing changes for other users or general rankings.
How do I embed Preferred Sources in a privacy-friendly way? ▾
With the deeplink to google.com/preferences/source?q=domain: no script, no third-party request on page load, no new consent category. The official button loads a Google script and belongs behind consent in most setups.
Who can be marked as a preferred source? ▾
Domains and subdomains — such as example.com or blog.example.com. Subdirectories like example.com/magazine/ cannot be marked separately. The feature is available for Top Stories globally in all languages of Google Search.
How strong is the effect? ▾
Unquantified. Google publishes no figures on the frequency of additional serving or the badge's effect on click behaviour. The only honest framing is loyalty infrastructure with an open effect coefficient — not a plannable traffic lever.