What Google Has in Store for SEOs in 2025: AI, Discover on Desktop, and New Content Requirements

In April 2025, the Google Search Central Live conference took place in Madrid. The event brought together leading Google experts, the SEO community, and professionals from around the world to discuss the future of search in the era of rapidly evolving artificial intelligence. Key topics included generative search, the role of structured data, content quality, and the expansion of Google Discover to desktop.

The Promodo team has prepared a report on how Google is adapting to these changes, what it plans to update in search results in the AI era, and how these updates will impact business SEO strategies. The article includes key insights, takeaways, and practical tips on how to maintain visibility and traffic in the new search landscape.

Google’s Mission in the AI Era

Despite emerging platforms, shifting user preferences, and the rise of AI technologies, Google’s top priority remains unchanged: providing maximum value to the user. Every change, including AI integration, is approached through the lens of enhancing the user experience.

The company aims to reassure the SEO community — especially amidst ongoing discussions about AI’s impact on search. Google representatives emphasized that core SEO elements like crawling and indexing are not going anywhere. Artificial intelligence isn’t here to replace existing tools or strategies in digital marketing, but rather to enhance them and unlock new opportunities.

SEO professionals are encouraged to leverage their current expertise while adapting to new technologies. There’s no need for radical strategic shifts—just thoughtful integration.

Generative Search and AI Overviews

AI Overviews (AIO) and other generative search capabilities were a major focus at the conference. Google representatives, including John Mueller, explained how LLMs (large language models) are being incorporated into the search ecosystem.

Key topics included:

• Retrieval-Augmented Generation (RAG): This technique allows LLMs to pull live data from Google’s index before generating a response. Unlike traditional language models that rely only on pre-trained data, RAG first retrieves relevant information (“retrieval”) and then creates a response based on it (“generation”). This hybrid model helps deliver more accurate, relevant, and up-to-date answers while reducing the risk of hallucinated facts.
  
  
• Grounding Mechanism: This ties generated answers to specific sources that validate the accuracy of the content. That’s why high-quality, structured, and fact-checked content on websites is more important than ever—it’s used for fact-checking in AIO responses. Websites that provide this kind of content are more likely to become grounding sources in generative search.

‍

Gain Benefits of Google Updates

with Promodo's SEO services!

Get In Touch

‍

Optimizing for AIO

Google emphasizes: there’s no need to invent new tactics just for AI Overviews. As users are still adapting to this new way of interacting with search, the best SEO strategy is to stick with tried-and-true optimization practices that have long proven effective in traditional search.
‍

AIO doesn’t replace SEO fundamentals — it actually reinforces their importance. High-quality content, clear site structure, solid technical optimization, and a user-friendly experience remain the foundation.

Control Over How Your Content Is Used in AI

Google gives webmasters control over how their content appears in generative features:

• To disable content from appearing in AI Overviews, you can use the no snippet directive in your robots.txt file.
• To prevent your content from being used to train Gemini or Vertex AI models, apply the Google-Extended token.
  ‍

🔒 Important: These settings do not affect your site’s indexing or rankings in traditional search results. They only determine whether Google can use your content in AI-generated answers or for model training.

AIO Analytics

Google has confirmed that impressions and clicks related to AI Overviews (AIO) are already included in the Performance Report in Google Search Console. However, there’s no separate AIO category — for now.

📊 The reason? AIO is still in an active testing phase, and splitting out this data too early could lead to confusion rather than clarity.

As for search activity in Gemini, that data is currently unavailable due to privacy and ethical considerations.
‍

Google Does Not Support LLMs.txt (Yet)

Within the SEO and tech communities, there’s been growing interest in establishing a universal standard that would allow website owners to clearly specify whether large language models (LLMs) are allowed to use their content.

One such idea is LLMs.txt — a file that, like robots.txt, would signal to AI platforms whether they’re allowed to access and use site content for training or generating responses.

While this could be a helpful tool for managing interaction with AI systems — including those from Google, OpenAI, Anthropic, and others — Google currently does not support this initiative. Instead, it continues to rely on existing crawling directives like robots.txt.
‍

AI-Generated Content Under the Microscope

Content quality was one of the key topics at the conference, especially in the context of the growing use of AI-generated texts. Google has officially clarified that if the main content of a page is created predominantly or entirely by AI — and lacks originality or added value — that page will receive the lowest rating according to the updated Quality Rater Guidelines.

‍

However, if AI is used merely as a tool — for example, for structuring, suggesting ideas, or editing — and the final result provides value and is created with human involvement, it will not be considered a violation.

Quality Rater Guidelines Update

• Google acknowledges generative AI as a potentially useful tool but warns against its misuse for creating spam or fake expertise.
• Spam is officially recognized as the creation of a large number of low-quality pages, often automated. AI is one of the main risks in this area.
• A new concept of “Filler” content has been introduced. This refers to content that visually takes up a lot of space on a page (such as long, generic introductions or inflated paragraphs) but doesn’t provide real value and prevents users from accessing useful, core content. Such content can lower a site's rating to Low, even if it's not harmful.
• Stronger requirements for E-E-A-T (Expertise, Authoritativeness, and Trustworthiness) and combating Misleading Claims: Google is tightening requirements for truthful information about authors and the site itself. Even exaggerated or "mildly misleading" claims about experience, expertise, or authority (such as inflated credentials or fabricated expertise) that aren’t outright lies, can now be sufficient grounds for a Low rating. Assessors are advised to rely on external reputation, verified facts, and the quality of core content, not just claims made on the site.
  ‍

Google expects transparency: who the author is, their experience, and whether the site truly deserves trust. These criteria directly affect rankings, especially in the era of generative search.

Structured Data Remains Relevant

Despite the rapid development of language models and their ability to "read" unstructured texts, Google emphasizes that structured data from Schema.org remains critically important.

Markup helps search engines interpret page content more accurately. It provides Google with clear signals about the content and relationships between elements on a page, making it easier to extract key information.

The types of structured data that are officially supported by Google have the most benefits. These can trigger rich snippets—visually noticeable elements directly in the search results (SERP) that attract attention and increase click-through rates (CTR).

Why do we need markup if we have LLMs? While large language models are impressive in understanding text, they still struggle with quickly and accurately extracting specific facts, such as product prices, event dates, or recipe ingredients.

This is where Schema.org markup acts as a "bridge" between content and machines. For AI tools, especially AIO, based on Retrieval-Augmented Generation (RAG), such data provides a reliable source of facts for "grounding" responses.

‍

Google also added that structured data on its own does not improve a website's search ranking. However, it helps the system better understand the page and activate rich snippets, which directly impact click-through rates (CTR) and engagement.

Big Announcement! Google Discover Expands Beyond Mobile

One of the most exciting announcements at the Google Search Central Live event in Madrid was the official confirmation that Google Discover will now be available on the desktop version of Google.
‍

‍

Exact launch date not yet announced, but this announcement highlights Google’s strategic focus on personalized content across all platforms.

What This Means for Publishers 

Google Discover has long been a powerful traffic channel for mobile audiences— in some niches, it can account for up to 60% of total traffic from Google. The arrival of Discover on desktop is expected to open a new window for reach, with a larger screen, different consumption patterns, and potentially higher engagement.

This is especially important for media, news platforms, analytics services, and anyone working with fresh, informative content.

How to Prepare Your Content for Desktop Discover Google has not announced any new or specific requirements for desktop, so all current recommendations remain valid:

• Publish quality, relevant, and unique content that aligns with your audience's interests.
• Use high-quality images (recommended width is at least 1200 px).
• Write honest and informative headlines, avoiding clickbait.
• Follow E-E-A-T principles.
• Add structured data to improve content understanding by Google if needed.
  
  

Conclusions 

2025 marks a new era in the evolution of Google Search— with the active implementation of generative AI, new content expectations, and expanded interfaces, including Google Discover on desktop. For SEO professionals, authors, and website owners, this means:

• Quality content is paramount. Artificial intelligence doesn’t replace the need for human contribution, originality, and accuracy.
  
  
• AI Overviews require precision. Structured data, verified sources, and clear information presentation are your main allies.
  
  
• No new "secret" tactics. Google advises sticking to classic SEO principles and avoiding shortcuts.
  
  
• Discover on desktop opens new opportunities. It’s important to start adapting your content to its demands—visual appeal, relevance, and E-E-A-T.
  
  

This is not about complication, but about a new level of quality. SEO in the age of AI remains about people, value, and trust.
‍

Gain Organic Traffic and Sales!

Message Promodo SEO experts to strengthen your SEO strategy.

Contact