Summary

• AI Overviews (AIOs) have reduced clickthrough rates. Our publishing clients have experienced significant losses from web search since Google released AI Overviews (AIOs). The generative summary experience has led to a -42% organic traffic decline across our portfolio.
• Breaking News is a notable exception. Breaking News is up +103% across all Google Surfaces, while evergreen content is down -35%. The Top Stories carousel, where Breaking News lives and provides publishers with the most valuable organic search real estate, has so far been protected from AI Overview incursion.
• Discover has driven breaking news growth. Although breaking news in Web Search has significantly outperformed other content types, Discover has been the real driving engine behind Breaking News growth across all surfaces. This mirrors the general trend we’re seeing, which is that as Web Search has declined, Discover has grown at a rapid pace. For the first time in our history, Discover and Web Search send an equal amount of traffic to the Define Media Group panel.
  

AI Overviews

When Google launched AI Overviews, publishers experienced immediate and unprecedented declines in organic search traffic.

Prior to the AIO launch, Define’s clients averaged a steady 1.7B organic search clicks over the previous 5 quarters (Q1 2023 – Q1 2024). After Google released AI Overviews, our portfolio experienced an immediate -16% search traffic loss, from which it never recovered. Traffic hovered around that level for a year as Google tested its AIO implementation.

In May 2025, Google entered an expansionary AIO phase. Since then, search clicks to our portfolio have declined each successive quarter vs. the previous. Compared to the pre-AIO average, organic search in Q4 2025 is down -42%.

Industry-Specific Impacts

AIO appearance rates are not uniform – it varies significantly across industries. This research from Ahrefs illustrates the variability:

AIOs have a 15% visibility rate for news content, nearly 3x lower than health and science-related information. We partner with major US national news providers, and have noticed that, for now, the Top Stories carousel has been one of the few protected corners against AIO incursion.

The Top Stories carousel is where breaking news lives. It requires up-to-the-minute updates, and the stakes for accurate information are higher than other verticals.

Currently, a news query like “iran war” doesn’t trigger an AIO. Instead, the Top Stories carousel appears, presenting a compelling image card and headline which encourages users to click through to the publisher’s story. AIOs are completely absent from this search result.

LLMs may be unable to retrieve and generate summaries at the pace required by breaking news events. And despite the progress they’ve made, the frontier models are still capable of hallucinations. Likely as a hedge against technical limitations and inaccurate information, breaking news content has been spared from generative summation.

Breaking News

To quantify our observational analysis, we undertook a large-scale content classification exercise, categorizing articles from the top 15 news sites in our panel into four segments: Breaking News, Evergreen, Landing Pages, and Homepage. Each of these segments were trended from November 2024 to date. The resulting data surprised us.

Across all Google Surfaces (Web Search, Discover, and Google News), Breaking News has grown +103% since November 2024, a growth rate exceeding our expectations. All other content types have declined.

Discover

As web search traffic has declined for publishers, the share of traffic from Discover has steadily increased.

Discover’s growth rate (+30%) has not increased as quickly as the Web Search decline rate (-43%) – and the growth began at a lower base – so net clicks from Google Surfaces to the portfolio are still down. However, it is notable that for the first time on record, our portfolio’s share of Web and Discover clicks have equalized.

Discover’s ascendancy as a traffic channel applies to breaking news as well. When we segment breaking news by Google Surface type, we see that Discover is responsible for nearly all of the breaking news growth:

Although Breaking News in Web Search has significantly outperformed other content types, holding steady at 2% while others are down over 30%, the clear takeaway from the data is that Discover is the real growth engine for breaking news.

Breaking News in Discover grew steadily through 2025, but experienced a step-wise increase after the December 2025 Core Update was released:

Some of those gains were given back after the February Discover Core Update, which was the first Discover-specific core update rolled out by Google. But Discover breaking news traffic is still up 2x compared to pre-December Core Update numbers.

Implications

1. Discover-Specific Optimization

Discover’s rise and web search’s fall should incentivize publishers to prioritize a Discover-specific content development strategy. The Discover platform was originally based on the Google search algorithm (and likely still shares many of its characteristics). But it’s also clear that Discover is maturing into its own unique system with dedicated signals independent of web search.

The first-ever Discover-specific Core Update, rumors of a soon-to-be-released Discover Profile, and the very slow rollout of Discover on desktop are evidence of this shift. Publishers need to scrutinize their Discover traffic patterns, to understand how to better optimize for Discover as a unique, stand-alone marketing and content channel.

Recent Google Discover optimization strategies from Harry Clarkson Bennett’s Leadership in SEO substack and Metehan Yeşilyurt’s Metehan.AI blog align with how we’re thinking about Discover-specific content approaches. Glenn Gabe and Lily Ray also provide excellent news and analysis around Discover and are worth following.

2. Evergreen vs. Breaking News Content Mix

Our expectation is that:

1. AI Overviews will continue to expand across all content verticals
2. Google will push AI Mode and conversational experiences into search results
3. Additional enhancements, like Web Guide, will continue to crowd out the classic link-based search result page

These and other unforeseen developments will likely trend towards increasing zero-click behavior. Evergreen content – and especially commoditized evergreen content – will continue to diminish as a viable traffic pathway from organic search. Publishers that have history and expertise as an evergreen content producer will need to evaluate the search viability of this content, and if they are able to, shift their resources and editorial mix towards breaking news and/or invest in non-organic marketing channels such as social and email.

Methodology

Content Classification

Google Search Console does not provide a “Top Stories” appearance filter. As a result, we looked at page performance over time to classify content according to the following criteria:

• Breaking News: Time-sensitive, event-driven content that generates a sharp traffic spike concentrated within one or two weeks. These pages cover developing stories, and their traffic decays rapidly once the news cycle moves on.
• Evergreen: Durable, reference-quality content that attracts steady traffic over many weeks or months. These pages serve ongoing informational needs (how-tos, explainers, product reviews, etc.) and show no single-week dominance in their traffic pattern.
• Landing Pages: Section fronts, topic hubs, and other navigational pages that sit between the homepage and individual articles.
• Homepage: The primary root URL (www or non-www) of a site.

Panel Composition

• Content Classification Analysis: Top 15 national and local news brands within the Define Media Group portfolio.
• Web and Discover Industry Trends: Define Media Group “Core” Sites – a 64-panel cohort of online businesses (publishers, ecommerce sites, data aggregators) within our network that represent general industry trends.

Data Sources

• Organic search metrics and dimensions (clicks, impressions, clickthrough rates, average position, pages, queries, etc.) from Google Search Console compiled at scale across the Define Media Group portfolio.

AI Usage

We used Claude for data analysis, content classification, and chart creation. We also used Gemini nano banana for the breaking news hero image. We chose not to use AI to write any portion of this blog.