Google will look beyond volume journalism

The Google ecosystem’s supply-demand dynamics are shifting faster than newsrooms can adapt.

For a decade, Google rewarded volume. Newsrooms chased a few trends, often drifting off-brand for traffic. The strategy: publish many “good enough” stories targeting single keywords. A few hits subsidized the misses; volume signaled authority.

Improved retrieval technology strains this model. Google now represents stories as semantic concepts (multi-vectors), searching for them cheaply — evidenced by Google’s MUVERA.

Features like Discover’s AI summary cards mainstream this by grouping stories into topics, summarizing facts, and picking a single “best-in-topic” story to showcase.

If trends continue, Google’s ecosystem may evolve as follows:

1. Topic clusters go mainstream. To cut LLM-generated clutter, expensive features like topic clustering, once restricted to niche products like Google News (since the 2010s), will go mainstream across Search, Gemini, Discover, and YouTube. Discover’s mid-2025 AI summary cards already signaled this shift.

Here’s how topic clusters look on YouTube.

2. On Discover, YouTube, and Gemini, traffic will go to fewer, deeper stories on a topic. Volume journalism will bleed traffic as AI Overviews summarize it. The total click pool will shrink; remaining clicks will favor stories too semantically deep to be easily summarized.

3. On Discover, YouTube, and Gemini, most traffic will go to “best-in-topic” stories with competitively more net-new information. Google will filter by geography, then identify “best-in-topic” stories across publishers using algorithms like Information Gain (a 2022 Google patent) to reward unique data.

To compete, publishers will invest in defensible news-gathering:

• Feet on the ground (reportage);
• Experts contributing knowledge, not just opinions;
• Data journalism;
• Analysis that is more sophisticated than what reasoning models can replicate.

This creates “verification latency.” RAG models view net-new information without consensus as potential hallucinations. Therefore, brand reputation and domain authority become “trust keys,” allowing high-authority domains to bypass consensus filters while others wait for validation.

4. Big media will invest in diverse topics, especially defensible niches where they can become the canonical source for Google. Large publishers will spread their bets, aiming to become the “canonical source” in specific topics in a certain language and geography to dominate the “best-in-topic” selection.

5. On Search’s lucrative “Top Stories” carousel, shallow recent stories for developing news will get prioritized. Google will likely continue to maintain two different models:

• Fast Feed: For breaking or developing news, Search will prefer shallow and fast for the Top Stories carousel. However, many of these are likely to get summarized.
• Deep Feed: Once the news cycle settles, Discover, YouTube, and Gemini will prefer fewer, deeper stories.

For publishers, balancing depth with freshness remains a delicate technical challenge.

6. Product teams will pivot before editorial. Workflow, incentives, and CMS architecture favor “publish and forget,” not “publish and maintain.”

Many news organizations already cluster stories by topics; a natural next step is to use LLMs to de-duplicate overlapping stories into a dense, auto-updating story.

Product teams will test such stories so on-platform users don’t have to find information across dozens of near-identical articles.

7. Publishers will need to figure out how to monetize attention. The programmatic display model is built on volume and refresh rates; the retrieval model is moving toward depth. This means that five users bouncing through two shallow pages (5-10 impressions) often generate more revenue than one user reading deeply for 10 minutes (2-3 impressions).

Some publishers must pivot to reader revenue/affiliates. Others will have to prove to direct-sold advertisers that one engaged reader is worth more than five users who bounce off easily.

8. Narrative strategy will bifurcate: radical voice vs. radical utility.

• High-voice UX: Focusing on distinct narrative craft (e.g., The Atlantic) where the experience of reading is the product, not just the facts. The goal is “irreducibility” — writing where the AI summary feels like a hollow substitute.
• High-utility UX: Standardized production formats (e.g., Axios’s Smart Brevity) optimize “time to value.” This also improves machine legibility and can make the publisher the preferred “structured data source” for AI models to cite.

Publishers will adopt AI editors as style linters to enforce this, rejecting drafts that are too generic to be voiced but too unstructured to be useful. Here’s one that I built for myself.

Publishers who try to do both under one domain will confuse the ranking algorithms and the users.

In 2026, human-written volume journalism will have diminishing returns. Publishers must bifurcate their business: a Fast Feed to capture breaking intent on Search, and a Deep Feed (Discover, YouTube, Gemini) where competitive “information gain” wins.