Loading article…
Google rolls Preferred Sources to 345,000 sites and introduces inline links and Gmail‑driven brand boosts in AI Overviews, reshaping visibility for publishers
Google has raised the number of Preferred Sources visible in AI Overviews to 345,000 — up from roughly 90,000 at the December rollout — and is adding more inline links and link context to its AI‑generated answers [1][3]. The changes give readers a clearer way to surface trusted sites, while giving publishers new levers to influence click‑through rates in AI Mode.
| At a glance | |
|---|---|
| Preferred Sources count | 345,000 (vs. ~90,000 in Dec) |
| New link format | Inline links plus end‑of‑answer links |
| Gmail signal impact | Email‑linked brands appear more often in AI Mode |
| CEO comment | AI Overview described as “more opinionated than it should be” |
Google’s Preferred Sources program, originally limited to Top Stories, now appears inside AI Overviews and AI Mode answers. Sites can prompt readers to add them, and the label is displayed whenever the site is cited [1]. The jump to 345,000 sources reflects rapid publisher adoption and signals that Google is treating reader loyalty as a visibility factor in AI answers. Industry observers note that a Preferred Source label can double click‑through likelihood, turning brand trust into a measurable traffic driver [1].
A separate test of Google’s Personal Intelligence feature shows that brands linked to a user’s Gmail data surface more frequently in AI Mode when the feature is enabled, making email the strongest signal for AI visibility [1]. Although the study’s sample size is small and the numbers are approximate, the finding suggests that email engagement could become a new SEO lever for brands seeking AI‑driven exposure.
In addition to Preferred Sources, Google announced five updates that reshape how links appear in AI Overviews. The updates add more inline links next to the supporting text, additional links at the bottom of responses, previews from public forums (e.g., Reddit), and desktop hover previews [3]. By moving citations closer to the relevant sentence, Google aims to improve the “click math” for cited pages, which previously clustered at the bottom and were often ignored.
The inclusion of forum previews introduces a new surface for content from discussion platforms, potentially surfacing brand mentions alongside AI‑generated answers. However, Google’s CEO Sundar Pichai recently called an AI Overview “more opinionated than it should be,” indicating that the company is still fine‑tuning the balance between answer generation and source attribution [1].
These moves underscore Google’s push to embed user‑curated trust signals and richer link context into AI‑driven search answers, blurring the line between traditional SERP listings and AI Overviews. The open question is how effectively the new signals will translate into measurable traffic gains for publishers and whether the “opinionated” AI responses will converge toward more neutral, source‑driven answers.
Coverage is mostly measured — 164 of 176 reports stay neutral.
Every Monday — the token unlocks, Fed dates & catalysts set to move crypto and markets this week. So you’re never blindsided.
Free · 3-min read · one-click unsubscribe
AI-assisted synthesis by the TrendWatcher Editorial Desk · sourced from 3 outlets · Jun 16, 2026 · How we report
According to the Seattle Times, the “AI fence” refers to Google’s effort to limit the amount of internet content its AI models can access, effectively creating a controlled environment for AI training.
Kotaku reports that the integration of Gemini into Google’s search has contributed to a decline of about 40% in human‑originated traffic to websites, a trend termed “Google Zero.”
Cloudflare’s data shows that AI‑training crawler requests comprised 52% of total crawler traffic as of June 2026, up from 22% in spring 2025.
Publishers who opt out of appearing in Google’s AI Overview summaries also risk being excluded from Google’s standard search indexing, potentially reducing their site visibility.