tikitaka.digital erreicht 98 / 100 · KI-bereit

Kategorie-Werte

Was du zuerst beheben solltest

• Machine-readable content density (signal-to-noise). Shrink the markup and inline JavaScript wrapped around your content so a larger share of the page an AI crawler fetches is actual readable text: server-render (or pre-render) the main content, trim heavy inline scripts and framework/hydration payloads, and keep the answer high in the HTML. AI crawlers and agents ingest the raw HTML and pay a context cost for noise — a leaner page is extracted and cited more reliably. This diagnoses the same gap that 'agent experience' delivery tools address, without cloaking or serving crawlers different content than people see.
• Direct quotations present (research-validated). Add direct, attributed quotations from credible people or sources to the page's main content (e.g. an expert quote, a customer testimonial, or a cited authority). Quotations are the strongest research-validated signal for being cited in AI answers (+27.8% in the KDD'24 GEO study). Use real, verifiable quotes — never fabricate attributions.
• Cited statistics present (research-validated). Add relevant quantitative statistics (percentages, counts, figures) and attribute each to a named, credible source (e.g. 'according to [source], NN% …'). Statistics with sources are a top research-validated GEO signal (+25.9%). Use real, verifiable numbers with real sources — never invent figures.
• Authoritative outbound citations (research-validated). Cite authoritative sources by linking out to reputable references (peer-reviewed research, official/government/edu sites, recognised industry sources) where you make factual claims. Cited sources are a research-validated GEO signal (+24.9%). Link only to genuinely relevant, credible sources.
• No keyword stuffing (research-validated anti-pattern). REDUCE keyword repetition — do NOT add more keywords. Rewrite over-optimised passages in natural language, replace repeated terms with synonyms and pronouns, and remove any keyword-dense lists/footers. Keyword stuffing is the worst-performing tactic in the KDD'24 GEO study and is discounted by AI engines; the goal here is lower density, not higher. Google's Search spam policy now explicitly counts 'attempting to manipulate generative AI responses in Google Search' as spam — keyword-dense text written for AI answers is a penalty risk, not an optimisation.
• Same content served to AI agents and browsers. Serve AI agents and human browsers the same content. Google's Search spam policy now explicitly counts 'attempting to manipulate generative AI responses in Google Search' as spam, and serving an AI agent a different page than a person is exactly that. If a bot gets less, your prerendering/edge rules are hiding content from AI (fix delivery); if a bot gets more or different content, that is cloaking — search engines penalise it and answer engines increasingly detect it. Diagnose with: curl -A "ChatGPT-User" vs a normal browser fetch of the same URL.