GEO / LLMO (answer-engine optimization)
Mitt verdikt
Strukturere innhold og entiteter slik at merkevarer blir sitert av ChatGPT, Google AI Overviews og Perplexity: schema.org, speakable, entitetsarkitektur og serverlogg-analyse av KI-crawlere. Målt mot reelle sitasjoner, ikke antatt.
Praktikerbrief
What we actually do
Entity architecture (schema.org with llmCard, FAQ, speakable blocks; Wikidata-backed entities), agent-discovery surfaces (llms.txt, .well-known agent cards), and server-log analysis of AI crawlers (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) to see what models can and cannot reach. The goal is being cited inside ChatGPT, Google AI Overviews, and Perplexity answers, not ranking a blue link.
Why Trial, not Adopt
Because we measure it on ourselves and the results are honest: our own monitoring baseline showed top citation presence across most models for one precise niche phrase - and zero presence on broad transactional queries where competitors hold third-party citations. On-page structure is necessary but not sufficient; the missing half is off-page corroboration (independent mentions models can verify). Any GEO pitch that skips measurement is selling assumptions.
When we reach for it
Brands whose buyers ask AI assistants before they ever search: we start with a fixed-scope AI visibility audit (crawler-log analysis, entity and schema gap review, live citation tests on priority queries), then decide whether ongoing optimization is worth it. Re-measurement against real citations is the exit criterion, not a report that ends the engagement.
Lagt til: 2026-07-07 · Sist gjennomgått: 2026-07-07