Eugen Ullrich — Blog https://eullrich.com/en/blog Insights on GEO, LLM visibility and technical SEO by Eugen Ullrich. en Why visibility in one AI engine says little about the next: one month of retrieval-to-citation data from eight B2B projects https://eullrich.com/en/blog/retrieval-citation-funnel https://eullrich.com/en/blog/retrieval-citation-funnel Sun, 19 Jul 2026 00:00:00 GMT I spent a month measuring how often ChatGPT, Perplexity and Google AI Overview actually cite a source they retrieved. Across eight B2B projects the pattern holds: three engines, three near-separate source pools. AI visibility is three targets, not one. Loop engineering: what the new way of working with AI agents is worth — and when a good prompt is enough https://eullrich.com/en/blog/loop-engineering-ai-agents https://eullrich.com/en/blog/loop-engineering-ai-agents Fri, 17 Jul 2026 00:00:00 GMT Loop engineering has AI agents work in cycles until a stop criterion takes hold. Which loop types exist, why loops fail in practice, and which seven questions to answer before your first own loop. A skill knows nothing: brand knowledge as an OKF bundle instead of a pile of files in the chat https://eullrich.com/en/blog/brand-knowledge-okf-bundle https://eullrich.com/en/blog/brand-knowledge-okf-bundle Wed, 15 Jul 2026 00:00:00 GMT Skills execute, they know nothing. How to store brand knowledge in the Open Knowledge Format: a folder of Markdown files any model reads without conversion. With my complete GEO bundle to download under CC BY 4.0. The prompt as a Lego figure: why AI visibility runs on entities, not keywords https://eullrich.com/en/blog/one-prompt-twelve-searches https://eullrich.com/en/blog/one-prompt-twelve-searches Sun, 05 Jul 2026 00:00:00 GMT AI systems break every prompt into several sub-queries and search for answers in parallel. Whoever optimizes only for the starting term is absent from the majority of these runs. Why entity mapping replaces the foundation that keyword thinking once was. llms.txt: what the file does, what Google says about it — and when you should create one https://eullrich.com/en/blog/llms-txt-guide https://eullrich.com/en/blog/llms-txt-guide Fri, 03 Jul 2026 00:00:00 GMT The complete guide to llms.txt: origin, structure, Google's official position, what the data shows about its use and who the file really pays off for — with my own analysis of 2,041 source domains, an example and practical steps. Not every rule belongs in the prompt: steering AI instructions and model memory https://eullrich.com/en/blog/not-every-rule-belongs-in-the-prompt https://eullrich.com/en/blog/not-every-rule-belongs-in-the-prompt Fri, 03 Jul 2026 00:00:00 GMT Six zones decide where an AI instruction belongs — from the product settings to the current prompt. How to instruct ChatGPT and Claude cleanly and keep their memory of you current. Google ends the June spam update in two days — and the forums report more spam, not less https://eullrich.com/en/blog/google-spam-update-june-2026 https://eullrich.com/en/blog/google-spam-update-june-2026 Mon, 29 Jun 2026 00:00:00 GMT The June 2026 spam update was over in two days. Why the loud complaints about traffic drops and rising spam are not yet a data picture, and what you should really check instead of forum panic. How ChatGPT chooses its sources — and the GEO levers that follow https://eullrich.com/en/blog/how-chatgpt-chooses-sources https://eullrich.com/en/blog/how-chatgpt-chooses-sources Mon, 29 Jun 2026 00:00:00 GMT A look at the traffic beneath the answer shows how ChatGPT gathers, selects and cites sources. What of it is actually evidenced, and where to act for your AI visibility. When the AI answers too shallowly: eight techniques to make it think deeper https://eullrich.com/en/blog/ai-techniques-deeper-thinking https://eullrich.com/en/blog/ai-techniques-deeper-thinking Sat, 27 Jun 2026 00:00:00 GMT Strong prompts are not magic formulas. They give the model a mode of thinking. Eight techniques to open up a topic and secure a decision — with ready-made prompt templates. Retrieved but not cited: the most valuable gap in your AI visibility https://eullrich.com/en/blog/retrieved-but-not-cited https://eullrich.com/en/blog/retrieved-but-not-cited Thu, 25 Jun 2026 00:00:00 GMT When an AI model retrieves your page but does not cite it, that is your biggest opportunity. How to measure the gap, prioritize it, and track it over time. The system prompt, taken from Claude Fable 5 — and which prompting techniques you can adopt from Anthropic https://eullrich.com/en/blog/system-prompt-claude-fable-5 https://eullrich.com/en/blog/system-prompt-claude-fable-5 Tue, 23 Jun 2026 00:00:00 GMT The complete 17,000-word system prompt was pulled from Claude Fable 5. What a system prompt even does, how far you can steer a model's answers — and eight techniques Anthropic builds into its models. GLM-5.2: the strongest open model is text-based and token-hungry https://eullrich.com/en/blog/glm-5-2-open-model-token-hunger https://eullrich.com/en/blog/glm-5-2-open-model-token-hunger Wed, 17 Jun 2026 00:00:00 GMT Z.ai released GLM-5.2 as open weights under an MIT license. In the independent Intelligence Index it leads the open field. Two caveats decide the practical value: it takes only text and consumes a strikingly high number of tokens. When your AI cuts corners: five techniques to get more out of every answer https://eullrich.com/en/blog/ai-techniques-sharper-answers https://eullrich.com/en/blog/ai-techniques-sharper-answers Fri, 12 Jun 2026 00:00:00 GMT Even a well-considered AI answer is often raw. Five techniques bring answers up to working quality, uncover hidden weaknesses in your ideas, and can be linked into chains. Where the tokens go: steering the context of AI models https://eullrich.com/en/blog/steering-ai-model-context https://eullrich.com/en/blog/steering-ai-model-context Mon, 01 Jun 2026 00:00:00 GMT The context window is finite, and the fuller it gets, the worse the model works. How to make the context visible, know its working zone and place a task into it without wasting quality and tokens. How Claude Code is built: memory, MCP, skills, subagents, hooks https://eullrich.com/en/blog/how-claude-code-is-built https://eullrich.com/en/blog/how-claude-code-is-built Sun, 10 May 2026 00:00:00 GMT Claude Code looks like a command line from the nineties. Beneath it lie six layers that make the tool a construction kit. A look under the hood, with examples from my work. Google Search Central APAC 2025: what really counts in SEO practice https://eullrich.com/en/blog/google-search-central-apac-2025 https://eullrich.com/en/blog/google-search-central-apac-2025 Tue, 12 Aug 2025 00:00:00 GMT A sober look back at Google Search Central Live APAC 2025 — the points that make a measurable difference in day-to-day SEO. Crawl budget: how HTTP status codes affect your SEO ranking https://eullrich.com/en/blog/crawl-budget-http-status-codes https://eullrich.com/en/blog/crawl-budget-http-status-codes Mon, 30 Jun 2025 00:00:00 GMT Why HTTP status codes govern crawl budget and how clean status-code management noticeably raises crawling efficiency. Why keyword research can distort your SEO strategy https://eullrich.com/en/blog/how-keyword-research-distorts-seo https://eullrich.com/en/blog/how-keyword-research-distorts-seo Sun, 18 May 2025 00:00:00 GMT Classic keyword research often optimizes for volume instead of value and thereby distorts the whole strategy. How to avoid it. Pricing first: why Germans watch prices https://eullrich.com/en/blog/pricing-first-german-market https://eullrich.com/en/blog/pricing-first-german-market Wed, 09 Apr 2025 00:00:00 GMT In the German market, price is often the entry point to the buying decision. What that means for dynamic pricing strategies, platform presence and value communication. The relevance of LLM visibility for online shops https://eullrich.com/en/blog/llm-visibility-for-online-shops https://eullrich.com/en/blog/llm-visibility-for-online-shops Fri, 21 Mar 2025 00:00:00 GMT Buying decisions are shifting into AI answers. Why LLM visibility becomes a channel of its own for online shops and how to occupy it.