Reading time: 11 minutes

Recap | The top 10 insights from SEO Day 2025

SEO isn't dead – AI just wants to play along. SEO Day 2025 in Cologne showed that those who rethink now will win. We (Radica and Emina) participated via live stream and summarised the most important insights for you. It's great that this option is available! It was an intense day full of insights, lectures and practical tips.
AI is changing the rules of the game. ChatGPT, Claude and Co. answer search queries directly. Google integrated AI mode into its search results some time ago, and traditional traffic sources are collapsing in some cases. But right now, new opportunities are emerging for anyone who is willing to rethink their approach. Two heads are better than one, which is why our marketing team split up during the conference to attend as many relevant sessions as possible. From GEO (Generative Engine Optimisation) to no-code automation to new link building strategies, the variety of topics was enormous.

In this article, we share our most important insights with you. No grey theory, just concrete insights that you can incorporate directly into your SEO strategy. Whether you're wondering how to optimise your content for AI systems, which tools really make your work easier, or why image links are suddenly more valuable than text links, you'll find the answers here.

Ready for the top 10 insights for your SEO work? Let's go!

Request a free initial consultation for your SEO advice now!

1. GEO is the new SEO

Eico Schweins summed up the new reality: People are increasingly asking AI instead of Google, and the lost Google traffic is unlikely to return. But SEO isn't dead; it has reached a new stage of evolution: GEO (Generative Engine Optimization).

What can we do? Optimize for RAG (Retrieval Augmented Generation). Here's how it works: AI asks Google -> Google responds -> AI passes the data on to us. And this is where the classic ranking factors come into play, because they are simpler and more cost-effective. Eico Schweins also named three GEO goals that we should all keep in mind. 

• Take advantage of first-mover advantage: Optimize now, before everyone else follows suit
• Strengthen your brand and thus be considered a notable source
• Increase traffic to be seen by AI

Takeaway: GEO is not science fiction, but a practical necessity. The good news: Much of it is “classic” good SEO, only viewed through AI glasses. Structured data, llms.txt, and “chunkable” content are now the basics for 2025.

2. Content must be “chunkable” and AI-readable.

Let's move straight on to chunkable content, which can be quickly understood by humans and optimally processed by AI. 

Experience, expertise, authoritativeness, and trustworthiness (EEAT) are essential. This means: show first-hand experience instead of just researching, demonstrate technical expertise, name experts as authors, and update content regularly. The inverted pyramid principle is a must—the most important information comes first. Every article needs a TL;DR at the beginning: 2-4 sentences that summarize the core message in a compact form. No more long-winded introductions, just valuable information right away.

Chunkable content means breaking information down into small chunks. Here is a brief overview:

Five key best practices:

• Clear heading hierarchy
• Short paragraphs (max. 3-4 sentences)
• Bullet points
• FAQ sections
• Sparse highlighting 

All of this makes content both readable for humans and quotable for AI. Clear, precise wording is essential. This is because AI has difficulty with nested sentences and ambiguity.

Takeaway: Content creation for AI is evolution, not revolution. Quality, expertise, and structure remain central.

3. AI replaces jobs – but also creates new ones

Marco Lauerwald presented facts in his talk “Skill gaps: upskilling and reskilling in marketing”: According to him, ChatGPT can perform over 50% of all human tasks. A test comparing humans and machines in 44 jobs showed that rental jobs (simple, repetitive tasks) are particularly at risk. In the field of SEO, ChatGPT was at least as good as humans, with Claude even winning most of the time. Only in image editing did AI perform significantly worse.

The JobFuturomat tool visualizes how many jobs can be rationalized away by AI. The parallel to the industrial revolution is obvious: old jobs are disappearing, new ones are being created, such as AI manager or prompt engineer. However, the real problem lies not with the technology, but with people: employees do not know how to use AI properly. Unfortunately, poor prompts lead to poor results.

Many people need to learn new skills. The EU AI Act will require everyone who works with AI to complete training. The OpenAI Academy already offers certifications. But the dilemma is that employees are unable to learn new things while working (too much to do – we all know that) and their own free time is important. This is how skill gaps arise. 

Takeaway: Lauerwald presented the »L17-Modell: Job« as a solution. This consists of 17 principles for energy-positive learning in a work context and offers the opportunity to close the resulting skill gaps.

4. B2B marketing needs personalization

“People will forget what you said. People will forget what you did. But people will never forget how you made them feel.” – Maya Angelou

This quote comes from Robin Heintze's presentation and was one of my personal highlights. His introduction: love letters. A love letter is clear, personal, emotional, and targeted. It conveys appreciation and makes the recipient feel good. This is exactly what is missing in many B2B communications – and this is exactly where marketing automation comes in.

The reality: most letters in the B2B environment seem unclear, impersonal, generic, or like spam. Messages without relevance don't stick. Successful automation, on the other hand, works like human-to-human communication. Data is used to create value, the target group is clearly defined, and the content addresses real desires, problems, and decision-making factors. Companies feel understood, and this is precisely what creates a need that we can then solve.

Five key best practices:

• Set up simple welcome sequences
  New subscribers receive a clearly structured sequence with relevant content such as white papers, webinars, or topic-specific emails. No self-congratulation, just content that matches the recipient's interests. Written like real messages from a person.
• Personalize everything consistently
  Content, tone, and offers are tailored to the behavior and needs of individual contacts.
• Strengthen lead generation with relevant content
  Webinars, case studies, and practical content build trust and lead qualified leads into automation.
• Use AI wisely
  AI supports scaling, personalization, and efficiency—without losing the human core of communication.
• Enrich data and qualify leads
  The CRM must be kept clean. Automated processes help to supplement data, identify interests, and develop leads in a targeted manner.

Takeaway: Marketing automation is not a technical setup, but rather automated relationship building. It works when recipients feel that you are genuinely interested in them. Communication remains human, personal, and relevant. Emails come from individuals, not teams or generic senders. This is the only way to be remembered and create a lasting impact.

5. N8N as a game changer for no-code automation

Matthias Uhlig's presentation, “SEO tools without programming knowledge: No-code solutions for your own needs,” provided a wealth of practical know-how. His solution? N8N – a central automation platform that can revolutionize the way we design SEO processes.

N8N offers several key advantages: for example, it can be hosted independently, which is essential from a data protection perspective, especially for German companies. The platform offers countless connections to virtually every relevant tool and is flexible. Particularly interesting: N8N works as the simplest MCP server (Model Context Protocol), which greatly simplifies integration with AI systems.
The practical applications are almost limitless. N8N can be linked to everything via APIs: Google Search Console, CRM systems, analytics tools, social media platforms, and much more. In addition, the platform can merge, analyze, and process data from various sources and make it available in any format.

Here is a concrete example: Instead of manually downloading Search Console data every week, importing it into Excel, comparing it with CRM data, and creating a PDF report, N8N automates this entire workflow. The strategic importance of N8N goes beyond saving time. In a world where SEO teams (and marketing teams) have to analyze more data, use more tools, and react faster, automation is no longer a luxury but a necessity.

Takeaway: N8N is the key to SEO automation without a team of developers. The platform combines flexibility, data protection (own hosting), and virtually unlimited integration options.

Request a free initial consultation for your SEO advice now!

6. Embeddings replace classic keyword thinking

The days when we primarily controlled SEO via individual keywords have changed. In one of the more technically oriented sessions at SEO Day, it was reported that search engines have been thinking in terms of concepts rather than keywords for some time now. Sebastian Erlhofer's session “Better SEO with embeddings – how to optimize with AI-generated vectors” was very mathematical and at times a little overwhelming for me ㋡. Those who don't come from a data science background will need to invest significantly more time to fully understand the concepts.
However, the principle of embeddings was very fascinating. Every word and every term is translated into a multidimensional vector. These vectors then map the meanings and relationships. Terms that are close to each other, for example, also have a small distance between them in the vector space. Similarity and semantic proximity, in turn, influence today's rankings even more than in the past. It is no longer a matter of placing a keyword as often as possible, but of covering a thematic concept holistically.

The session also mentioned the Screaming Frog tool as a central working tool and Python as a programming language. Admittedly, technical know-how is required here, because without basic knowledge of Python, it is difficult to use the full power of embeddings.

Takeaway: We need to think much more in terms of semantic clusters rather than isolated keywords—even if the technical implementation still presents a steep learning curve for me at the moment.

7. Backlinks remain highly relevant – image links beat text links

Despite artificial intelligence, backlinks are still an important ranking factor. Julian Strote's talk “Backlinks in the age of AI – Why it has rarely been easier to rank with links” was interesting in many ways. It began with a brief history of the early days of links. Keywords such as Alta Vista, Yahoo web catalog, and Page Rank were mentioned, and I had to quickly calculate how old I really am ㋡.

But back to backlinks. 

The quote from John Müller of Google in the talk summed it up: “Create a brand and make sure you are mentioned on other websites.” It is no longer about classic follow links with perfect anchor texts. Search engines now recognize and evaluate brand mentions. Tools such as Otterly help to track and evaluate these mentions. Where are people talking about us? In what context does our brand appear? All these signals are factored into the ranking. In addition, according to recent studies, image links have a higher value than pure text links. Visual content that is linked is superior in its effect to classic text links (this includes infographics or diagrams).

It was very interesting to learn that platforms such as Quora and Reddit play a fairly significant role in the link building universe, but are often neglected. It is particularly interesting that English-language Quora answers can easily link to German websites—and according to Ahrefs studies, this works surprisingly well.

Takeaway: Link building is becoming more visual and contextual. Instead of producing masses of text links, we should invest in shareable visual content and position our brand where authentic conversations take place.

8. Listicles rank highly

Google, and presumably AI systems too, love structures, and for good reason. They are scannable for users, clear for crawlers, ideal for featured snippets, and perfect for the GEO era. But not every list ranks automatically Among other things, the substance of each individual list item and a good headline such as “The 7 best ...” or “10 reasons why ...” are important here. If the whole thing is enriched with visual elements – perfect. By the way: don't forget anchor links, as users like to jump directly to their preferred topic. (Good thing we didn't forget the anchor links in our blog post ㋡.)

My Takeaway: the listicle format is still relevant today and has become more professional – if you're going to use listicles, make sure they have depth, data, and real expertise.

9. Internal linking with a maximum crawl depth of 3

Laura Piccolomo summed it up perfectly in her talk “SEO like the Mafia – using internal linking to create an organized structure.” Every website needs not only a well-thought-out hierarchy, but also good internal linking. Google does not crawl every page with the same frequency. Through intelligent internal linking, we can signal to Google which web pages are important and thus save crawl budget. Ideally, every important web page should be accessible from the home page with a maximum of three clicks. I know this is sometimes difficult, but it is feasible. Not least because the right SEO strategy can help here. I found the presentation of the SEO tool Screaming Frog and its practical application as a strategic consultant particularly exciting. Even though the overall presentation was more of a refresher for me personally, I found the more detailed description of Screaming Frog very exciting.

Takeaway: Treat your website structure like a well-organized family. Every page has its role, every link has its meaning. 

10. PBN – The dark side of link building

Private blog networks (PBNs) are the shadow economy of SEO. In Daniel Bruckhaus' session “PBNs in the age of AI – dead or stronger than ever?”, it became clear that link building remains highly relevant, but the methods have certainly changed. PBNs are networks of websites that serve exclusively to place backlinks on money sites. There are entire marketplaces that offer backlinks for as little as $400 per link. This practice is particularly widespread in the travel industry, but also in the pharmaceutical and real estate sectors. In his talk, Daniel Bruckhaus showed us concrete examples of these soulless websites. Minimal investment in design, little to no real user interaction, and content that appears very generic and is AI-generated. Sounds kind of creepy.

But the part about detecting backlinks was really interesting. Some SEO tools are often blocked, so they can't find any backlinks at all. Therefore, one tool is not enough; a multi-tool analysis is a must. Although Google is getting better at recognizing unnatural link patterns, PBNs are evolving in parallel.

Takeaway: PBNs are real. In my opinion, this approach is risky for serious SEO.

Request a free initial consultation for your SEO advice now!

References

[1] AI in everyday life at digital agencies
[2] Starfruit AI